re-fixed baselines from ErrorPrediction to ClassificationError
This commit is contained in:
Frank Seide
Родитель
320f2f53b1
Коммит
abe66249e7
|
|
@ -303,7 +303,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
AutoName217 = Times()
|
||||
LossFunction = CrossEntropyWithSoftmax()
|
||||
PredictionError = ErrorPrediction()
|
||||
PredictionError = ClassificationError()
|
||||
|
||||
Validating network. 36 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -342,7 +342,7 @@ Validating --> AutoName216 = Sigmoid (AutoName215) : [2048 x 1 x *8] -> [2048 x
|
|||
Validating --> AutoName217 = Times (AutoName185, AutoName216) : [9304 x 2048], [2048 x 1 x *8] -> [9304 x 1 x *8]
|
||||
Validating --> AutoName218 = InputValue() : -> [9304 x 1 x *8]
|
||||
Validating --> LossFunction = CrossEntropyWithSoftmax (AutoName218, AutoName217) : [9304 x 1 x *8], [9304 x 1 x *8] -> [1]
|
||||
Validating --> PredictionError = ErrorPrediction (AutoName218, AutoName217) : [9304 x 1 x *8], [9304 x 1 x *8] -> [1]
|
||||
Validating --> PredictionError = ClassificationError (AutoName218, AutoName217) : [9304 x 1 x *8], [9304 x 1 x *8] -> [1]
|
||||
|
||||
Validating network. 21 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -405,7 +405,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
AutoName382 = Times()
|
||||
LossFunction = CrossEntropyWithSoftmax()
|
||||
PredictionError = ErrorPrediction()
|
||||
PredictionError = ClassificationError()
|
||||
|
||||
Validating network. 36 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -444,7 +444,7 @@ Validating --> AutoName381 = Sigmoid (AutoName380) : [2048 x 1 x *9] -> [2048 x
|
|||
Validating --> AutoName382 = Times (AutoName350, AutoName381) : [9304 x 2048], [2048 x 1 x *9] -> [9304 x 1 x *9]
|
||||
Validating --> AutoName383 = InputValue() : -> [9304 x 1 x *9]
|
||||
Validating --> LossFunction = CrossEntropyWithSoftmax (AutoName383, AutoName382) : [9304 x 1 x *9], [9304 x 1 x *9] -> [1]
|
||||
Validating --> PredictionError = ErrorPrediction (AutoName383, AutoName382) : [9304 x 1 x *9], [9304 x 1 x *9] -> [1]
|
||||
Validating --> PredictionError = ClassificationError (AutoName383, AutoName382) : [9304 x 1 x *9], [9304 x 1 x *9] -> [1]
|
||||
|
||||
Validating network. 21 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -865,7 +865,7 @@ Post-processing network...
|
|||
AutoName1611 = LearnableParameter()
|
||||
AutoName1665 = Plus()
|
||||
lossFunction = CrossEntropyWithSoftmax()
|
||||
predictionError = ErrorPrediction()
|
||||
predictionError = ClassificationError()
|
||||
|
||||
Loop[0] --> Loop_AutoName1497 -> 35 nodes
|
||||
|
||||
|
|
@ -1182,7 +1182,7 @@ Validating --> AutoName1664 = LearnableParameter() : -> [9304 x 1]
|
|||
Validating --> AutoName1665 = Plus (AutoName1663, AutoName1664) : [9304 x 1 x *16], [9304 x 1] -> [9304 x 1 x *16]
|
||||
Validating --> AutoName1666 = InputValue() : -> [9304 x 1 x *16]
|
||||
Validating --> lossFunction = CrossEntropyWithSoftmax (AutoName1666, AutoName1665) : [9304 x 1 x *16], [9304 x 1 x *16] -> [1]
|
||||
Validating --> predictionError = ErrorPrediction (AutoName1666, AutoName1665) : [9304 x 1 x *16], [9304 x 1 x *16] -> [1]
|
||||
Validating --> predictionError = ClassificationError (AutoName1666, AutoName1665) : [9304 x 1 x *16], [9304 x 1 x *16] -> [1]
|
||||
|
||||
Validating network. 180 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -1591,7 +1591,7 @@ Post-processing network...
|
|||
AutoName5595 = LearnableParameter()
|
||||
AutoName5649 = Plus()
|
||||
lossFunction = CrossEntropyWithSoftmax()
|
||||
predictionError = ErrorPrediction()
|
||||
predictionError = ClassificationError()
|
||||
|
||||
Loop[0] --> Loop_AutoName5481 -> 35 nodes
|
||||
|
||||
|
|
@ -1908,7 +1908,7 @@ Validating --> AutoName5648 = LearnableParameter() : -> [9304 x 1]
|
|||
Validating --> AutoName5649 = Plus (AutoName5647, AutoName5648) : [9304 x 1 x *17], [9304 x 1] -> [9304 x 1 x *17]
|
||||
Validating --> AutoName5650 = InputValue() : -> [9304 x 1 x *17]
|
||||
Validating --> lossFunction = CrossEntropyWithSoftmax (AutoName5650, AutoName5649) : [9304 x 1 x *17], [9304 x 1 x *17] -> [1]
|
||||
Validating --> predictionError = ErrorPrediction (AutoName5650, AutoName5649) : [9304 x 1 x *17], [9304 x 1 x *17] -> [1]
|
||||
Validating --> predictionError = ClassificationError (AutoName5650, AutoName5649) : [9304 x 1 x *17], [9304 x 1 x *17] -> [1]
|
||||
|
||||
Validating network. 180 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -79,8 +79,8 @@ labelDim = 10
|
|||
h1 = DNNSigmoidLayer (featDim, hiddenDim, featScaled, 1)
|
||||
z = DNNLayer (hiddenDim, labelDim, h1, 1)
|
||||
ce = CrossEntropyWithSoftmax (labels, z)
|
||||
errs = ErrorPrediction (labels, z)
|
||||
top5Errs = ErrorPrediction (labels, z, topN=5)
|
||||
errs = ClassificationError (labels, z)
|
||||
top5Errs = ClassificationError (labels, z, topN=5)
|
||||
featureNodes = (features)
|
||||
labelNodes = (labels)
|
||||
criterionNodes = (ce)
|
||||
|
|
@ -171,8 +171,8 @@ labelDim = 10
|
|||
h1 = DNNSigmoidLayer (featDim, hiddenDim, featScaled, 1)
|
||||
z = DNNLayer (hiddenDim, labelDim, h1, 1)
|
||||
ce = CrossEntropyWithSoftmax (labels, z)
|
||||
errs = ErrorPrediction (labels, z)
|
||||
top5Errs = ErrorPrediction (labels, z, topN=5)
|
||||
errs = ClassificationError (labels, z)
|
||||
top5Errs = ClassificationError (labels, z, topN=5)
|
||||
featureNodes = (features)
|
||||
labelNodes = (labels)
|
||||
criterionNodes = (ce)
|
||||
|
|
@ -287,8 +287,8 @@ labelDim = 10
|
|||
h1 = DNNSigmoidLayer (featDim, hiddenDim, featScaled, 1)
|
||||
z = DNNLayer (hiddenDim, labelDim, h1, 1)
|
||||
ce = CrossEntropyWithSoftmax (labels, z)
|
||||
errs = ErrorPrediction (labels, z)
|
||||
top5Errs = ErrorPrediction (labels, z, topN=5)
|
||||
errs = ClassificationError (labels, z)
|
||||
top5Errs = ClassificationError (labels, z, topN=5)
|
||||
featureNodes = (features)
|
||||
labelNodes = (labels)
|
||||
criterionNodes = (ce)
|
||||
|
|
@ -360,9 +360,9 @@ Post-processing network...
|
|||
|
||||
4 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
ol.z = Plus()
|
||||
top5Errs = ErrorPrediction()
|
||||
top5Errs = ClassificationError()
|
||||
|
||||
Validating network. 17 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -380,9 +380,9 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 200], [200 x 1 x *] -> [10 x 1
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x 1 x *], [10 x 1] -> [10 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> unnamed89 = LearnableParameter() : -> [1 x 1]
|
||||
Validating --> top5Errs = ErrorPrediction (labels, ol.z, unnamed89) : [10 x *], [10 x 1 x *], [1 x 1] -> [1]
|
||||
Validating --> top5Errs = ClassificationError (labels, ol.z, unnamed89) : [10 x *], [10 x 1 x *], [1 x 1] -> [1]
|
||||
|
||||
Validating network. 9 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -401,8 +401,8 @@ Post-processing network complete.
|
|||
08/16/2016 10:49:44: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:49:44: Evaluation criterion node(s):
|
||||
08/16/2016 10:49:44: top5Errs = ErrorPrediction
|
||||
08/16/2016 10:49:44: errs = ErrorPrediction
|
||||
08/16/2016 10:49:44: top5Errs = ClassificationError
|
||||
08/16/2016 10:49:44: errs = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -475,8 +475,8 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
top5Errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
top5Errs = ClassificationError()
|
||||
|
||||
Validating network. 17 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -494,9 +494,9 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 200], [200 x 1 x *1] -> [10 x 1
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x 1 x *1], [10 x 1] -> [10 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> unnamed89 = LearnableParameter() : -> [1 x 1]
|
||||
Validating --> top5Errs = ErrorPrediction (labels, ol.z, unnamed89) : [10 x *1], [10 x 1 x *1], [1 x 1] -> [1]
|
||||
Validating --> top5Errs = ClassificationError (labels, ol.z, unnamed89) : [10 x *1], [10 x 1 x *1], [1 x 1] -> [1]
|
||||
|
||||
Validating network. 9 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -79,8 +79,8 @@ labelDim = 10
|
|||
h1 = DNNSigmoidLayer (featDim, hiddenDim, featScaled, 1)
|
||||
z = DNNLayer (hiddenDim, labelDim, h1, 1)
|
||||
ce = CrossEntropyWithSoftmax (labels, z)
|
||||
errs = ErrorPrediction (labels, z)
|
||||
top5Errs = ErrorPrediction (labels, z, topN=5)
|
||||
errs = ClassificationError (labels, z)
|
||||
top5Errs = ClassificationError (labels, z, topN=5)
|
||||
featureNodes = (features)
|
||||
labelNodes = (labels)
|
||||
criterionNodes = (ce)
|
||||
|
|
@ -171,8 +171,8 @@ labelDim = 10
|
|||
h1 = DNNSigmoidLayer (featDim, hiddenDim, featScaled, 1)
|
||||
z = DNNLayer (hiddenDim, labelDim, h1, 1)
|
||||
ce = CrossEntropyWithSoftmax (labels, z)
|
||||
errs = ErrorPrediction (labels, z)
|
||||
top5Errs = ErrorPrediction (labels, z, topN=5)
|
||||
errs = ClassificationError (labels, z)
|
||||
top5Errs = ClassificationError (labels, z, topN=5)
|
||||
featureNodes = (features)
|
||||
labelNodes = (labels)
|
||||
criterionNodes = (ce)
|
||||
|
|
@ -287,8 +287,8 @@ labelDim = 10
|
|||
h1 = DNNSigmoidLayer (featDim, hiddenDim, featScaled, 1)
|
||||
z = DNNLayer (hiddenDim, labelDim, h1, 1)
|
||||
ce = CrossEntropyWithSoftmax (labels, z)
|
||||
errs = ErrorPrediction (labels, z)
|
||||
top5Errs = ErrorPrediction (labels, z, topN=5)
|
||||
errs = ClassificationError (labels, z)
|
||||
top5Errs = ClassificationError (labels, z, topN=5)
|
||||
featureNodes = (features)
|
||||
labelNodes = (labels)
|
||||
criterionNodes = (ce)
|
||||
|
|
@ -360,9 +360,9 @@ Post-processing network...
|
|||
|
||||
4 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
ol.z = Plus()
|
||||
top5Errs = ErrorPrediction()
|
||||
top5Errs = ClassificationError()
|
||||
|
||||
Validating network. 17 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -380,9 +380,9 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 200], [200 x 1 x *] -> [10 x 1
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x 1 x *], [10 x 1] -> [10 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> unnamed89 = LearnableParameter() : -> [1 x 1]
|
||||
Validating --> top5Errs = ErrorPrediction (labels, ol.z, unnamed89) : [10 x *], [10 x 1 x *], [1 x 1] -> [1]
|
||||
Validating --> top5Errs = ClassificationError (labels, ol.z, unnamed89) : [10 x *], [10 x 1 x *], [1 x 1] -> [1]
|
||||
|
||||
Validating network. 9 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -401,8 +401,8 @@ Post-processing network complete.
|
|||
08/16/2016 03:00:47: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:00:47: Evaluation criterion node(s):
|
||||
08/16/2016 03:00:47: top5Errs = ErrorPrediction
|
||||
08/16/2016 03:00:47: errs = ErrorPrediction
|
||||
08/16/2016 03:00:47: top5Errs = ClassificationError
|
||||
08/16/2016 03:00:47: errs = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -475,8 +475,8 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
top5Errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
top5Errs = ClassificationError()
|
||||
|
||||
Validating network. 17 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -494,9 +494,9 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 200], [200 x 1 x *1] -> [10 x 1
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x 1 x *1], [10 x 1] -> [10 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> unnamed89 = LearnableParameter() : -> [1 x 1]
|
||||
Validating --> top5Errs = ErrorPrediction (labels, ol.z, unnamed89) : [10 x *1], [10 x 1 x *1], [1 x 1] -> [1]
|
||||
Validating --> top5Errs = ClassificationError (labels, ol.z, unnamed89) : [10 x *1], [10 x 1 x *1], [1 x 1] -> [1]
|
||||
|
||||
Validating network. 9 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -306,7 +306,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
ol.z = Plus()
|
||||
|
||||
Validating network. 27 nodes to process in pass 1.
|
||||
|
|
@ -337,7 +337,7 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 128], [128 x 1 x *] -> [10 x 1
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x 1 x *], [10 x 1] -> [10 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
|
||||
Validating network. 16 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -364,7 +364,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:49:52: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:49:52: Evaluation criterion node(s):
|
||||
08/16/2016 10:49:52: errs = ErrorPrediction
|
||||
08/16/2016 10:49:52: errs = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -460,7 +460,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
ol.z = Plus()
|
||||
|
||||
Validating network. 27 nodes to process in pass 1.
|
||||
|
|
@ -491,7 +491,7 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 128], [128 x 1 x *1] -> [10 x 1
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x 1 x *1], [10 x 1] -> [10 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
|
||||
Validating network. 16 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -306,7 +306,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
ol.z = Plus()
|
||||
|
||||
Validating network. 27 nodes to process in pass 1.
|
||||
|
|
@ -337,7 +337,7 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 128], [128 x 1 x *] -> [10 x 1
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x 1 x *], [10 x 1] -> [10 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
|
||||
Validating network. 16 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -364,7 +364,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:01:08: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:01:08: Evaluation criterion node(s):
|
||||
08/16/2016 03:01:08: errs = ErrorPrediction
|
||||
08/16/2016 03:01:08: errs = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -460,7 +460,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
ol.z = Plus()
|
||||
|
||||
Validating network. 27 nodes to process in pass 1.
|
||||
|
|
@ -491,7 +491,7 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 128], [128 x 1 x *1] -> [10 x 1
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x 1 x *1], [10 x 1] -> [10 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
|
||||
Validating network. 16 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -330,7 +330,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
ol.z = Plus()
|
||||
|
||||
Validating network. 36 nodes to process in pass 1.
|
||||
|
|
@ -372,7 +372,7 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 128], [128 x *] -> [10 x *]
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x *], [10 x 1] -> [10 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
|
||||
Validating network. 16 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -405,7 +405,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:50:07: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:50:07: Evaluation criterion node(s):
|
||||
08/16/2016 10:50:07: errs = ErrorPrediction
|
||||
08/16/2016 10:50:07: errs = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -504,7 +504,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
ol.z = Plus()
|
||||
|
||||
Validating network. 36 nodes to process in pass 1.
|
||||
|
|
@ -544,7 +544,7 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 128], [128 x *1] -> [10 x *1]
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x *1], [10 x 1] -> [10 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
|
||||
Validating network. 16 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -330,7 +330,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
ol.z = Plus()
|
||||
|
||||
Validating network. 36 nodes to process in pass 1.
|
||||
|
|
@ -372,7 +372,7 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 128], [128 x *] -> [10 x *]
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x *], [10 x 1] -> [10 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
|
||||
Validating network. 16 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -405,7 +405,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:01:37: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:01:37: Evaluation criterion node(s):
|
||||
08/16/2016 03:01:37: errs = ErrorPrediction
|
||||
08/16/2016 03:01:37: errs = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -504,7 +504,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
errs = ErrorPrediction()
|
||||
errs = ClassificationError()
|
||||
ol.z = Plus()
|
||||
|
||||
Validating network. 36 nodes to process in pass 1.
|
||||
|
|
@ -544,7 +544,7 @@ Validating --> ol.t = Times (ol.W, h1.y) : [10 x 128], [128 x *1] -> [10 x *1]
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol.z = Plus (ol.t, ol.b) : [10 x *1], [10 x 1] -> [10 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ErrorPrediction (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> errs = ClassificationError (labels, ol.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
|
||||
Validating network. 16 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -322,7 +322,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 34 nodes to process in pass 1.
|
||||
|
|
@ -360,7 +360,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, h1_d) : [10 x 64], [64 x 1
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x 1 x *], [10] -> [10 x 1 x *]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
|
||||
Validating network. 21 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -391,7 +391,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:50:38: CE = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:50:38: Evaluation criterion node(s):
|
||||
08/16/2016 10:50:38: Err = ErrorPrediction
|
||||
08/16/2016 10:50:38: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -544,7 +544,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 34 nodes to process in pass 1.
|
||||
|
|
@ -582,7 +582,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, h1_d) : [10 x 64], [64 x 1
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x 1 x *1], [10] -> [10 x 1 x *1]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
|
||||
Validating network. 21 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -322,7 +322,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 34 nodes to process in pass 1.
|
||||
|
|
@ -360,7 +360,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, h1_d) : [10 x 64], [64 x 1
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x 1 x *], [10] -> [10 x 1 x *]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *], [10 x 1 x *] -> [1]
|
||||
|
||||
Validating network. 21 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -391,7 +391,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:02:09: CE = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:02:09: Evaluation criterion node(s):
|
||||
08/16/2016 03:02:09: Err = ErrorPrediction
|
||||
08/16/2016 03:02:09: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -544,7 +544,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 34 nodes to process in pass 1.
|
||||
|
|
@ -582,7 +582,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, h1_d) : [10 x 64], [64 x 1
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x 1 x *1], [10] -> [10 x 1 x *1]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
|
||||
Validating network. 21 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -349,7 +349,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 45 nodes to process in pass 1.
|
||||
|
|
@ -398,7 +398,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, h1.y) : [10 x 64], [64 x *]
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *], [10] -> [10 x *]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
|
||||
Validating network. 20 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -437,7 +437,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:50:46: CE = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:50:46: Evaluation criterion node(s):
|
||||
08/16/2016 10:50:46: Err = ErrorPrediction
|
||||
08/16/2016 10:50:46: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -553,7 +553,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 45 nodes to process in pass 1.
|
||||
|
|
@ -602,7 +602,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, h1.y) : [10 x 64], [64 x *1
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *1], [10] -> [10 x *1]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
|
||||
Validating network. 20 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -349,7 +349,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 45 nodes to process in pass 1.
|
||||
|
|
@ -398,7 +398,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, h1.y) : [10 x 64], [64 x *]
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *], [10] -> [10 x *]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
|
||||
Validating network. 20 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -437,7 +437,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:02:27: CE = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:02:27: Evaluation criterion node(s):
|
||||
08/16/2016 03:02:27: Err = ErrorPrediction
|
||||
08/16/2016 03:02:27: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -553,7 +553,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 45 nodes to process in pass 1.
|
||||
|
|
@ -602,7 +602,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, h1.y) : [10 x 64], [64 x *1
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *1], [10] -> [10 x *1]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
|
||||
Validating network. 20 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -587,7 +587,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 184 nodes to process in pass 1.
|
||||
|
|
@ -777,7 +777,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, pool) : [10 x 1 x 1 x 64],
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *], [10] -> [10 x *]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
|
||||
Validating network. 75 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -882,7 +882,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:50:54: CE = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:50:54: Evaluation criterion node(s):
|
||||
08/16/2016 10:50:54: Err = ErrorPrediction
|
||||
08/16/2016 10:50:54: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1148,7 +1148,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 184 nodes to process in pass 1.
|
||||
|
|
@ -1336,7 +1336,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, pool) : [10 x 1 x 1 x 64],
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *1], [10] -> [10 x *1]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
|
||||
Validating network. 75 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -587,7 +587,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 184 nodes to process in pass 1.
|
||||
|
|
@ -777,7 +777,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, pool) : [10 x 1 x 1 x 64],
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *], [10] -> [10 x *]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
|
||||
Validating network. 75 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -882,7 +882,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:02:44: CE = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:02:44: Evaluation criterion node(s):
|
||||
08/16/2016 03:02:44: Err = ErrorPrediction
|
||||
08/16/2016 03:02:44: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1148,7 +1148,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 184 nodes to process in pass 1.
|
||||
|
|
@ -1336,7 +1336,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, pool) : [10 x 1 x 1 x 64],
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *1], [10] -> [10 x *1]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
|
||||
Validating network. 75 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -1481,7 +1481,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 949 nodes to process in pass 1.
|
||||
|
|
@ -2436,7 +2436,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, pool) : [10 x 1 x 1 x 64],
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *], [10] -> [10 x *]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
|
||||
Validating network. 390 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -2901,7 +2901,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:51:04: CE = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:51:04: Evaluation criterion node(s):
|
||||
08/16/2016 10:51:04: Err = ErrorPrediction
|
||||
08/16/2016 10:51:04: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -4149,7 +4149,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 949 nodes to process in pass 1.
|
||||
|
|
@ -5102,7 +5102,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, pool) : [10 x 1 x 1 x 64],
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *1], [10] -> [10 x *1]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
|
||||
Validating network. 390 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -1481,7 +1481,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 949 nodes to process in pass 1.
|
||||
|
|
@ -2436,7 +2436,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, pool) : [10 x 1 x 1 x 64],
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *], [10] -> [10 x *]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
|
||||
Validating network. 390 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -2901,7 +2901,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:03:11: CE = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:03:11: Evaluation criterion node(s):
|
||||
08/16/2016 03:03:11: Err = ErrorPrediction
|
||||
08/16/2016 03:03:11: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -4149,7 +4149,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 949 nodes to process in pass 1.
|
||||
|
|
@ -5102,7 +5102,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, pool) : [10 x 1 x 1 x 64],
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *1], [10] -> [10 x *1]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
|
||||
Validating network. 390 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -316,7 +316,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 32 nodes to process in pass 1.
|
||||
|
|
@ -352,7 +352,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, conv4.y) : [10 x 7 x 7 x 32
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *], [10] -> [10 x *]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
|
||||
Validating network. 19 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -383,7 +383,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:51:24: CE = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:51:24: Evaluation criterion node(s):
|
||||
08/16/2016 10:51:24: Err = ErrorPrediction
|
||||
08/16/2016 10:51:24: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -492,7 +492,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 32 nodes to process in pass 1.
|
||||
|
|
@ -528,7 +528,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, conv4.y) : [10 x 7 x 7 x 32
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *1], [10] -> [10 x *1]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
|
||||
Validating network. 19 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -316,7 +316,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 32 nodes to process in pass 1.
|
||||
|
|
@ -352,7 +352,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, conv4.y) : [10 x 7 x 7 x 32
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *], [10] -> [10 x *]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *], [10 x *] -> [1]
|
||||
|
||||
Validating network. 19 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -383,7 +383,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:03:59: CE = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:03:59: Evaluation criterion node(s):
|
||||
08/16/2016 03:03:59: Err = ErrorPrediction
|
||||
08/16/2016 03:03:59: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -492,7 +492,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
CE = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
OutputNodes.z = Plus()
|
||||
|
||||
Validating network. 32 nodes to process in pass 1.
|
||||
|
|
@ -528,7 +528,7 @@ Validating --> OutputNodes.t = Times (OutputNodes.W, conv4.y) : [10 x 7 x 7 x 32
|
|||
Validating --> OutputNodes.b = LearnableParameter() : -> [10]
|
||||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [10 x *1], [10] -> [10 x *1]
|
||||
Validating --> CE = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, OutputNodes.z) : [10 x *1], [10 x *1] -> [1]
|
||||
|
||||
Validating network. 19 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -81,7 +81,7 @@ Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -182,7 +182,7 @@ Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -315,7 +315,7 @@ configparameters: Multigpu.cntk:Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -395,7 +395,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -424,7 +424,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *] -> [2 x 1 x *]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *], [2 x 1] -> [2 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *] -> [2 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [2 x *] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -447,7 +447,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:01:27: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:01:27: Evaluation criterion node(s):
|
||||
08/16/2016 10:01:27: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 10:01:27: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -499,141 +499,141 @@ BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sampl
|
|||
BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sample 0), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:01:27: Starting minibatch loop.
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.69846765 * 250; EvalErrorPrediction = 0.50400000 * 250; time = 0.0208s; samplesPerSecond = 12032.5
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76129944 * 250; EvalErrorPrediction = 0.46400000 * 250; time = 0.0104s; samplesPerSecond = 24029.2
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72963208 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0263s; samplesPerSecond = 9510.0
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.74041528 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0289s; samplesPerSecond = 8665.2
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70611035 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0197s; samplesPerSecond = 12660.8
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74740723 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0259s; samplesPerSecond = 9634.3
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75085840 * 250; EvalErrorPrediction = 0.40400000 * 250; time = 0.0103s; samplesPerSecond = 24163.9
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78210742 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0168s; samplesPerSecond = 14848.3
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70286572 * 250; EvalErrorPrediction = 0.47200000 * 250; time = 0.0170s; samplesPerSecond = 14742.3
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69580322 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0292s; samplesPerSecond = 8552.3
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70703613 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0334s; samplesPerSecond = 7480.3
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74512988 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0104s; samplesPerSecond = 23941.8
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70837598 * 250; EvalErrorPrediction = 0.47600000 * 250; time = 0.0166s; samplesPerSecond = 15043.0
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69913086 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0166s; samplesPerSecond = 15038.5
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70321875 * 250; EvalErrorPrediction = 0.53600000 * 250; time = 0.0206s; samplesPerSecond = 12148.9
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69290918 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0260s; samplesPerSecond = 9610.2
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74415527 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0107s; samplesPerSecond = 23353.6
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73745117 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0166s; samplesPerSecond = 15081.1
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71849609 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0168s; samplesPerSecond = 14905.8
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71476953 * 250; EvalErrorPrediction = 0.50400000 * 250; time = 0.0136s; samplesPerSecond = 18331.1
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69918457 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0290s; samplesPerSecond = 8620.1
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69749512 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0107s; samplesPerSecond = 23454.4
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70658887 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0104s; samplesPerSecond = 23973.9
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69760742 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0199s; samplesPerSecond = 12538.9
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69499219 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0135s; samplesPerSecond = 18504.8
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69291211 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0293s; samplesPerSecond = 8538.8
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70718945 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0305s; samplesPerSecond = 8199.1
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69039453 * 250; EvalErrorPrediction = 0.46800000 * 250; time = 0.0169s; samplesPerSecond = 14832.4
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70257422 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0167s; samplesPerSecond = 14931.6
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71058984 * 250; EvalErrorPrediction = 0.42800000 * 250; time = 0.0166s; samplesPerSecond = 15085.7
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69296875 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0167s; samplesPerSecond = 14995.2
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69641211 * 250; EvalErrorPrediction = 0.51600000 * 250; time = 0.0168s; samplesPerSecond = 14916.5
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69531055 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0329s; samplesPerSecond = 7601.3
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69090430 * 250; EvalErrorPrediction = 0.47600000 * 250; time = 0.0208s; samplesPerSecond = 12036.6
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.68339063 * 250; EvalErrorPrediction = 0.48000000 * 250; time = 0.0168s; samplesPerSecond = 14893.4
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.67383984 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0330s; samplesPerSecond = 7576.2
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.65904102 * 250; EvalErrorPrediction = 0.26400000 * 250; time = 0.0104s; samplesPerSecond = 24010.8
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.64259766 * 250; EvalErrorPrediction = 0.36000000 * 250; time = 0.0135s; samplesPerSecond = 18487.0
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.60433398 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0167s; samplesPerSecond = 15004.2
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.56497070 * 250; EvalErrorPrediction = 0.11600000 * 250; time = 0.0136s; samplesPerSecond = 18390.5
|
||||
08/16/2016 10:01:28: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70222344 * 10000; EvalErrorPrediction = 0.46820000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.776535s
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.69846765 * 250; EvalClassificationError = 0.50400000 * 250; time = 0.0208s; samplesPerSecond = 12032.5
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76129944 * 250; EvalClassificationError = 0.46400000 * 250; time = 0.0104s; samplesPerSecond = 24029.2
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72963208 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0263s; samplesPerSecond = 9510.0
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.74041528 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0289s; samplesPerSecond = 8665.2
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70611035 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0197s; samplesPerSecond = 12660.8
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74740723 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0259s; samplesPerSecond = 9634.3
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75085840 * 250; EvalClassificationError = 0.40400000 * 250; time = 0.0103s; samplesPerSecond = 24163.9
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78210742 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0168s; samplesPerSecond = 14848.3
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70286572 * 250; EvalClassificationError = 0.47200000 * 250; time = 0.0170s; samplesPerSecond = 14742.3
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69580322 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0292s; samplesPerSecond = 8552.3
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70703613 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0334s; samplesPerSecond = 7480.3
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74512988 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0104s; samplesPerSecond = 23941.8
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70837598 * 250; EvalClassificationError = 0.47600000 * 250; time = 0.0166s; samplesPerSecond = 15043.0
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69913086 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0166s; samplesPerSecond = 15038.5
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70321875 * 250; EvalClassificationError = 0.53600000 * 250; time = 0.0206s; samplesPerSecond = 12148.9
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69290918 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0260s; samplesPerSecond = 9610.2
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74415527 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0107s; samplesPerSecond = 23353.6
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73745117 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0166s; samplesPerSecond = 15081.1
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71849609 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0168s; samplesPerSecond = 14905.8
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71476953 * 250; EvalClassificationError = 0.50400000 * 250; time = 0.0136s; samplesPerSecond = 18331.1
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69918457 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0290s; samplesPerSecond = 8620.1
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69749512 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0107s; samplesPerSecond = 23454.4
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70658887 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0104s; samplesPerSecond = 23973.9
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69760742 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0199s; samplesPerSecond = 12538.9
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69499219 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0135s; samplesPerSecond = 18504.8
|
||||
08/16/2016 10:01:27: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69291211 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0293s; samplesPerSecond = 8538.8
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70718945 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0305s; samplesPerSecond = 8199.1
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69039453 * 250; EvalClassificationError = 0.46800000 * 250; time = 0.0169s; samplesPerSecond = 14832.4
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70257422 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0167s; samplesPerSecond = 14931.6
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71058984 * 250; EvalClassificationError = 0.42800000 * 250; time = 0.0166s; samplesPerSecond = 15085.7
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69296875 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0167s; samplesPerSecond = 14995.2
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69641211 * 250; EvalClassificationError = 0.51600000 * 250; time = 0.0168s; samplesPerSecond = 14916.5
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69531055 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0329s; samplesPerSecond = 7601.3
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69090430 * 250; EvalClassificationError = 0.47600000 * 250; time = 0.0208s; samplesPerSecond = 12036.6
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.68339063 * 250; EvalClassificationError = 0.48000000 * 250; time = 0.0168s; samplesPerSecond = 14893.4
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.67383984 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0330s; samplesPerSecond = 7576.2
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.65904102 * 250; EvalClassificationError = 0.26400000 * 250; time = 0.0104s; samplesPerSecond = 24010.8
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.64259766 * 250; EvalClassificationError = 0.36000000 * 250; time = 0.0135s; samplesPerSecond = 18487.0
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.60433398 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0167s; samplesPerSecond = 15004.2
|
||||
08/16/2016 10:01:28: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.56497070 * 250; EvalClassificationError = 0.11600000 * 250; time = 0.0136s; samplesPerSecond = 18390.5
|
||||
08/16/2016 10:01:28: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70222344 * 10000; EvalClassificationError = 0.46820000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.776535s
|
||||
08/16/2016 10:01:28: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100054.995555/Examples/Other/Simple2d_MultiGpu@release_cpu/Models/multigpu.dnn.1'
|
||||
|
||||
08/16/2016 10:01:28: Starting Epoch 2: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 1: frames [10000..20000] (first sequence at sample 10000), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:01:28: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1).
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.50722371 * 250; EvalErrorPrediction = 0.14800000 * 250; time = 0.0397s; samplesPerSecond = 6295.5
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.45786101 * 250; EvalErrorPrediction = 0.12800000 * 250; time = 0.0285s; samplesPerSecond = 8776.9
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.37902995 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0178s; samplesPerSecond = 14020.5
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.34590577 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0176s; samplesPerSecond = 14178.0
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.29942918 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0174s; samplesPerSecond = 14344.7
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.28291648 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0301s; samplesPerSecond = 8297.1
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.25680062 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0530s; samplesPerSecond = 4715.7
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.21806843 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0388s; samplesPerSecond = 6450.9
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.22671616 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0293s; samplesPerSecond = 8533.6
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.20709374 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0177s; samplesPerSecond = 14159.5
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.18895447 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0365s; samplesPerSecond = 6855.7
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.17506560 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0441s; samplesPerSecond = 5669.8
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.18710038 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0252s; samplesPerSecond = 9901.0
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.18230681 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0205s; samplesPerSecond = 12218.4
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.18466931 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0175s; samplesPerSecond = 14290.6
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.17889979 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0174s; samplesPerSecond = 14329.9
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.18170165 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0215s; samplesPerSecond = 11627.4
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21059295 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0206s; samplesPerSecond = 12147.1
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.16428288 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0301s; samplesPerSecond = 8297.9
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17104948 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0500s; samplesPerSecond = 5002.3
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13190985 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0206s; samplesPerSecond = 12160.7
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17235489 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0174s; samplesPerSecond = 14329.1
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12426324 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0205s; samplesPerSecond = 12183.2
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21852627 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0177s; samplesPerSecond = 14104.4
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21640896 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0487s; samplesPerSecond = 5133.5
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.17959436 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0157s; samplesPerSecond = 15952.0
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16189965 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0145s; samplesPerSecond = 17266.4
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13475075 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0175s; samplesPerSecond = 14282.4
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16423768 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0333s; samplesPerSecond = 7510.0
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14635259 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0298s; samplesPerSecond = 8393.5
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.14974090 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0174s; samplesPerSecond = 14368.6
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12504713 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0175s; samplesPerSecond = 14289.0
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16433451 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0371s; samplesPerSecond = 6744.0
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14200378 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0221s; samplesPerSecond = 11319.9
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13708748 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0208s; samplesPerSecond = 12010.0
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.13991044 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0149s; samplesPerSecond = 16734.7
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15786864 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0208s; samplesPerSecond = 12029.1
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16220493 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0155s; samplesPerSecond = 16121.8
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13517917 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0199s; samplesPerSecond = 12571.7
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15440438 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0217s; samplesPerSecond = 11501.1
|
||||
08/16/2016 10:01:29: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.20309370 * 10000; EvalErrorPrediction = 0.08040000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=1.02227s
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.50722371 * 250; EvalClassificationError = 0.14800000 * 250; time = 0.0397s; samplesPerSecond = 6295.5
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.45786101 * 250; EvalClassificationError = 0.12800000 * 250; time = 0.0285s; samplesPerSecond = 8776.9
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.37902995 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0178s; samplesPerSecond = 14020.5
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.34590577 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0176s; samplesPerSecond = 14178.0
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.29942918 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0174s; samplesPerSecond = 14344.7
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.28291648 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0301s; samplesPerSecond = 8297.1
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.25680062 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0530s; samplesPerSecond = 4715.7
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.21806843 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0388s; samplesPerSecond = 6450.9
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.22671616 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0293s; samplesPerSecond = 8533.6
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.20709374 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0177s; samplesPerSecond = 14159.5
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.18895447 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0365s; samplesPerSecond = 6855.7
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.17506560 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0441s; samplesPerSecond = 5669.8
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.18710038 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0252s; samplesPerSecond = 9901.0
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.18230681 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0205s; samplesPerSecond = 12218.4
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.18466931 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0175s; samplesPerSecond = 14290.6
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.17889979 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0174s; samplesPerSecond = 14329.9
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.18170165 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0215s; samplesPerSecond = 11627.4
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21059295 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0206s; samplesPerSecond = 12147.1
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.16428288 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0301s; samplesPerSecond = 8297.9
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17104948 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0500s; samplesPerSecond = 5002.3
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13190985 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0206s; samplesPerSecond = 12160.7
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17235489 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0174s; samplesPerSecond = 14329.1
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12426324 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0205s; samplesPerSecond = 12183.2
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21852627 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0177s; samplesPerSecond = 14104.4
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21640896 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0487s; samplesPerSecond = 5133.5
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.17959436 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0157s; samplesPerSecond = 15952.0
|
||||
08/16/2016 10:01:28: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16189965 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0145s; samplesPerSecond = 17266.4
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13475075 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0175s; samplesPerSecond = 14282.4
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16423768 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0333s; samplesPerSecond = 7510.0
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14635259 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0298s; samplesPerSecond = 8393.5
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.14974090 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0174s; samplesPerSecond = 14368.6
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12504713 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0175s; samplesPerSecond = 14289.0
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16433451 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0371s; samplesPerSecond = 6744.0
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14200378 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0221s; samplesPerSecond = 11319.9
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13708748 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0208s; samplesPerSecond = 12010.0
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.13991044 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0149s; samplesPerSecond = 16734.7
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15786864 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0208s; samplesPerSecond = 12029.1
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16220493 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0155s; samplesPerSecond = 16121.8
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13517917 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0199s; samplesPerSecond = 12571.7
|
||||
08/16/2016 10:01:29: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15440438 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0217s; samplesPerSecond = 11501.1
|
||||
08/16/2016 10:01:29: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.20309370 * 10000; EvalClassificationError = 0.08040000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=1.02227s
|
||||
08/16/2016 10:01:29: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100054.995555/Examples/Other/Simple2d_MultiGpu@release_cpu/Models/multigpu.dnn.2'
|
||||
|
||||
08/16/2016 10:01:29: Starting Epoch 3: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 2: frames [20000..30000] (first sequence at sample 20000), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:01:29: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1).
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18478506 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0216s; samplesPerSecond = 11585.3
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12741733 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0184s; samplesPerSecond = 13576.6
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17535235 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0235s; samplesPerSecond = 10656.9
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14042800 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0150s; samplesPerSecond = 16696.7
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16643002 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0247s; samplesPerSecond = 10109.6
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19327050 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0389s; samplesPerSecond = 6424.8
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12260149 * 250; EvalErrorPrediction = 0.04000000 * 250; time = 0.0292s; samplesPerSecond = 8568.7
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16504305 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0328s; samplesPerSecond = 7631.0
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12425912 * 250; EvalErrorPrediction = 0.04000000 * 250; time = 0.0315s; samplesPerSecond = 7945.3
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19996755 * 250; EvalErrorPrediction = 0.10800000 * 250; time = 0.0943s; samplesPerSecond = 2649.9
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14253075 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0703s; samplesPerSecond = 3554.8
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12335900 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0494s; samplesPerSecond = 5064.0
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16695660 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0353s; samplesPerSecond = 7090.2
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19907855 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0346s; samplesPerSecond = 7225.4
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.16895044 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0346s; samplesPerSecond = 7233.4
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13285834 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0376s; samplesPerSecond = 6645.0
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14406293 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0176s; samplesPerSecond = 14231.2
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20987060 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0171s; samplesPerSecond = 14639.6
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19265041 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0389s; samplesPerSecond = 6432.9
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.15040079 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0277s; samplesPerSecond = 9019.4
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15551715 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0205s; samplesPerSecond = 12207.0
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13682837 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0212s; samplesPerSecond = 11784.1
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17235013 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0174s; samplesPerSecond = 14356.3
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14431340 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0176s; samplesPerSecond = 14196.5
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13791050 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0184s; samplesPerSecond = 13580.3
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14160704 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0175s; samplesPerSecond = 14275.1
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16921888 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0172s; samplesPerSecond = 14549.3
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18580557 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0206s; samplesPerSecond = 12133.6
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16487179 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0157s; samplesPerSecond = 15918.5
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15450410 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0204s; samplesPerSecond = 12249.5
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18731137 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0540s; samplesPerSecond = 4628.8
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13205502 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0113s; samplesPerSecond = 22137.6
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14591704 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0174s; samplesPerSecond = 14338.2
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13912720 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0175s; samplesPerSecond = 14267.0
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20110201 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0199s; samplesPerSecond = 12535.1
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12560399 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0245s; samplesPerSecond = 10196.2
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18609894 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0184s; samplesPerSecond = 13563.4
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15309858 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0174s; samplesPerSecond = 14405.9
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11872821 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0175s; samplesPerSecond = 14303.7
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.12948843 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0178s; samplesPerSecond = 14041.0
|
||||
08/16/2016 10:01:30: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15779327 * 10000; EvalErrorPrediction = 0.07250000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=1.10281s
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18478506 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0216s; samplesPerSecond = 11585.3
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12741733 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0184s; samplesPerSecond = 13576.6
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17535235 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0235s; samplesPerSecond = 10656.9
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14042800 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0150s; samplesPerSecond = 16696.7
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16643002 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0247s; samplesPerSecond = 10109.6
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19327050 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0389s; samplesPerSecond = 6424.8
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12260149 * 250; EvalClassificationError = 0.04000000 * 250; time = 0.0292s; samplesPerSecond = 8568.7
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16504305 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0328s; samplesPerSecond = 7631.0
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12425912 * 250; EvalClassificationError = 0.04000000 * 250; time = 0.0315s; samplesPerSecond = 7945.3
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19996755 * 250; EvalClassificationError = 0.10800000 * 250; time = 0.0943s; samplesPerSecond = 2649.9
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14253075 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0703s; samplesPerSecond = 3554.8
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12335900 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0494s; samplesPerSecond = 5064.0
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16695660 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0353s; samplesPerSecond = 7090.2
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19907855 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0346s; samplesPerSecond = 7225.4
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.16895044 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0346s; samplesPerSecond = 7233.4
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13285834 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0376s; samplesPerSecond = 6645.0
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14406293 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0176s; samplesPerSecond = 14231.2
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20987060 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0171s; samplesPerSecond = 14639.6
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19265041 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0389s; samplesPerSecond = 6432.9
|
||||
08/16/2016 10:01:29: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.15040079 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0277s; samplesPerSecond = 9019.4
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15551715 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0205s; samplesPerSecond = 12207.0
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13682837 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0212s; samplesPerSecond = 11784.1
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17235013 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0174s; samplesPerSecond = 14356.3
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14431340 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0176s; samplesPerSecond = 14196.5
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13791050 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0184s; samplesPerSecond = 13580.3
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14160704 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0175s; samplesPerSecond = 14275.1
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16921888 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0172s; samplesPerSecond = 14549.3
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18580557 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0206s; samplesPerSecond = 12133.6
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16487179 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0157s; samplesPerSecond = 15918.5
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15450410 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0204s; samplesPerSecond = 12249.5
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18731137 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0540s; samplesPerSecond = 4628.8
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13205502 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0113s; samplesPerSecond = 22137.6
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14591704 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0174s; samplesPerSecond = 14338.2
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13912720 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0175s; samplesPerSecond = 14267.0
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20110201 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0199s; samplesPerSecond = 12535.1
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12560399 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0245s; samplesPerSecond = 10196.2
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18609894 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0184s; samplesPerSecond = 13563.4
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15309858 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0174s; samplesPerSecond = 14405.9
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11872821 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0175s; samplesPerSecond = 14303.7
|
||||
08/16/2016 10:01:30: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.12948843 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0178s; samplesPerSecond = 14041.0
|
||||
08/16/2016 10:01:30: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15779327 * 10000; EvalClassificationError = 0.07250000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=1.10281s
|
||||
08/16/2016 10:01:30: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100054.995555/Examples/Other/Simple2d_MultiGpu@release_cpu/Models/multigpu.dnn'
|
||||
08/16/2016 10:01:30: CNTKCommandTrainEnd: Multigpu_Demo_Train
|
||||
|
||||
|
|
@ -651,7 +651,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -680,7 +680,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *1] -> [2 x 1 x *1]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *1], [2 x 1] -> [2 x 1 x *1]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *1] -> [2 x 1 x *1]
|
||||
Validating --> Prior = Mean (labels) : [2 x *1] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -709,8 +709,8 @@ Memory Sharing: Out of 25 matrices, 2 are shared as 1, and 23 are not shared.
|
|||
|
||||
BlockRandomizer::StartEpoch: epoch 0: frames [0..603] (first sequence at sample 0), data subset 0 of 1
|
||||
Actual gradient aggregation time: 0.000192
|
||||
08/16/2016 10:01:30: Minibatch[1-1]: EvalErrorPrediction = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10665885 * 603
|
||||
08/16/2016 10:01:30: Final Results: Minibatch[1-1]: EvalErrorPrediction = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10665885 * 603; perplexity = 1.11255464
|
||||
08/16/2016 10:01:30: Minibatch[1-1]: EvalClassificationError = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10665885 * 603
|
||||
08/16/2016 10:01:30: Final Results: Minibatch[1-1]: EvalClassificationError = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10665885 * 603; perplexity = 1.11255464
|
||||
|
||||
08/16/2016 10:01:30: Action "test" complete.
|
||||
|
||||
|
|
|
|||
|
|
@ -81,7 +81,7 @@ Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -182,7 +182,7 @@ Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -315,7 +315,7 @@ configparameters: Multigpu.cntk:Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -396,7 +396,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -425,7 +425,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *] -> [2 x 1 x *]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *], [2 x 1] -> [2 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *] -> [2 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [2 x *] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -448,7 +448,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:01:32: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:01:32: Evaluation criterion node(s):
|
||||
08/16/2016 10:01:32: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 10:01:32: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -500,141 +500,141 @@ BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sampl
|
|||
BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sample 0), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:01:32: Starting minibatch loop.
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70124231 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0076s; samplesPerSecond = 32761.1
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76372424 * 250; EvalErrorPrediction = 0.46400000 * 250; time = 0.0062s; samplesPerSecond = 40374.7
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72703027 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0061s; samplesPerSecond = 40836.3
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73895923 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0061s; samplesPerSecond = 41077.9
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70621924 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0061s; samplesPerSecond = 41010.5
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74767041 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0061s; samplesPerSecond = 41308.7
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75094434 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0061s; samplesPerSecond = 40690.1
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78058936 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0061s; samplesPerSecond = 40990.3
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70407129 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0061s; samplesPerSecond = 40763.1
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69555762 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0061s; samplesPerSecond = 41247.3
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70626123 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0061s; samplesPerSecond = 40976.9
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74540430 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0061s; samplesPerSecond = 41179.4
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70824414 * 250; EvalErrorPrediction = 0.47600000 * 250; time = 0.0060s; samplesPerSecond = 41480.0
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69895020 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0060s; samplesPerSecond = 41397.6
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70353223 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0061s; samplesPerSecond = 40763.1
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69346387 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0061s; samplesPerSecond = 41186.2
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74449902 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0062s; samplesPerSecond = 40643.8
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73767969 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0060s; samplesPerSecond = 41820.0
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71876855 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0060s; samplesPerSecond = 41862.0
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71509473 * 250; EvalErrorPrediction = 0.50400000 * 250; time = 0.0061s; samplesPerSecond = 41138.7
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69956152 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0059s; samplesPerSecond = 42108.8
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0059s; samplesPerSecond = 42337.0
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70736035 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0059s; samplesPerSecond = 42030.9
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69820508 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0059s; samplesPerSecond = 42430.4
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69537109 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0059s; samplesPerSecond = 42286.9
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69347266 * 250; EvalErrorPrediction = 0.51200000 * 250; time = 0.0059s; samplesPerSecond = 42387.2
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70801172 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0060s; samplesPerSecond = 41652.8
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69131641 * 250; EvalErrorPrediction = 0.46800000 * 250; time = 0.0059s; samplesPerSecond = 42294.0
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70370312 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0058s; samplesPerSecond = 42771.6
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71200195 * 250; EvalErrorPrediction = 0.43600000 * 250; time = 0.0058s; samplesPerSecond = 42808.2
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69506836 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0058s; samplesPerSecond = 42800.9
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69935352 * 250; EvalErrorPrediction = 0.51600000 * 250; time = 0.0058s; samplesPerSecond = 43305.0
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69887109 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0058s; samplesPerSecond = 42764.3
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69604492 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0058s; samplesPerSecond = 43110.9
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.69011719 * 250; EvalErrorPrediction = 0.48800000 * 250; time = 0.0065s; samplesPerSecond = 38302.4
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.68419531 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0058s; samplesPerSecond = 43148.1
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.67551367 * 250; EvalErrorPrediction = 0.32400000 * 250; time = 0.0059s; samplesPerSecond = 42094.6
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.67028516 * 250; EvalErrorPrediction = 0.40000000 * 250; time = 0.0059s; samplesPerSecond = 42294.0
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.65152734 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0059s; samplesPerSecond = 42329.8
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.63594727 * 250; EvalErrorPrediction = 0.22000000 * 250; time = 0.0060s; samplesPerSecond = 41666.7
|
||||
08/16/2016 10:01:33: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70729233 * 10000; EvalErrorPrediction = 0.47740000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.245257s
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70124231 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0076s; samplesPerSecond = 32761.1
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76372424 * 250; EvalClassificationError = 0.46400000 * 250; time = 0.0062s; samplesPerSecond = 40374.7
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72703027 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0061s; samplesPerSecond = 40836.3
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73895923 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0061s; samplesPerSecond = 41077.9
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70621924 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0061s; samplesPerSecond = 41010.5
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74767041 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0061s; samplesPerSecond = 41308.7
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75094434 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0061s; samplesPerSecond = 40690.1
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78058936 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0061s; samplesPerSecond = 40990.3
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70407129 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0061s; samplesPerSecond = 40763.1
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69555762 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0061s; samplesPerSecond = 41247.3
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70626123 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0061s; samplesPerSecond = 40976.9
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74540430 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0061s; samplesPerSecond = 41179.4
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70824414 * 250; EvalClassificationError = 0.47600000 * 250; time = 0.0060s; samplesPerSecond = 41480.0
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69895020 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0060s; samplesPerSecond = 41397.6
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70353223 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0061s; samplesPerSecond = 40763.1
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69346387 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0061s; samplesPerSecond = 41186.2
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74449902 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0062s; samplesPerSecond = 40643.8
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73767969 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0060s; samplesPerSecond = 41820.0
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71876855 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0060s; samplesPerSecond = 41862.0
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71509473 * 250; EvalClassificationError = 0.50400000 * 250; time = 0.0061s; samplesPerSecond = 41138.7
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69956152 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0059s; samplesPerSecond = 42108.8
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0059s; samplesPerSecond = 42337.0
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70736035 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0059s; samplesPerSecond = 42030.9
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69820508 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0059s; samplesPerSecond = 42430.4
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69537109 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0059s; samplesPerSecond = 42286.9
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69347266 * 250; EvalClassificationError = 0.51200000 * 250; time = 0.0059s; samplesPerSecond = 42387.2
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70801172 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0060s; samplesPerSecond = 41652.8
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69131641 * 250; EvalClassificationError = 0.46800000 * 250; time = 0.0059s; samplesPerSecond = 42294.0
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70370312 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0058s; samplesPerSecond = 42771.6
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71200195 * 250; EvalClassificationError = 0.43600000 * 250; time = 0.0058s; samplesPerSecond = 42808.2
|
||||
08/16/2016 10:01:32: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69506836 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0058s; samplesPerSecond = 42800.9
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69935352 * 250; EvalClassificationError = 0.51600000 * 250; time = 0.0058s; samplesPerSecond = 43305.0
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69887109 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0058s; samplesPerSecond = 42764.3
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69604492 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0058s; samplesPerSecond = 43110.9
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.69011719 * 250; EvalClassificationError = 0.48800000 * 250; time = 0.0065s; samplesPerSecond = 38302.4
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.68419531 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0058s; samplesPerSecond = 43148.1
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.67551367 * 250; EvalClassificationError = 0.32400000 * 250; time = 0.0059s; samplesPerSecond = 42094.6
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.67028516 * 250; EvalClassificationError = 0.40000000 * 250; time = 0.0059s; samplesPerSecond = 42294.0
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.65152734 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0059s; samplesPerSecond = 42329.8
|
||||
08/16/2016 10:01:33: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.63594727 * 250; EvalClassificationError = 0.22000000 * 250; time = 0.0060s; samplesPerSecond = 41666.7
|
||||
08/16/2016 10:01:33: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70729233 * 10000; EvalClassificationError = 0.47740000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.245257s
|
||||
08/16/2016 10:01:33: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100054.995555/Examples/Other/Simple2d_MultiGpu@release_gpu/Models/multigpu.dnn.1'
|
||||
|
||||
08/16/2016 10:01:33: Starting Epoch 2: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 1: frames [10000..20000] (first sequence at sample 10000), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:01:33: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1).
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.61550018 * 250; EvalErrorPrediction = 0.27600000 * 250; time = 0.0108s; samplesPerSecond = 23111.8
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.59409242 * 250; EvalErrorPrediction = 0.28800000 * 250; time = 0.0094s; samplesPerSecond = 26612.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.53884306 * 250; EvalErrorPrediction = 0.20400000 * 250; time = 0.0093s; samplesPerSecond = 26890.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.52450125 * 250; EvalErrorPrediction = 0.15200000 * 250; time = 0.0093s; samplesPerSecond = 26942.6
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.49237463 * 250; EvalErrorPrediction = 0.16400000 * 250; time = 0.0092s; samplesPerSecond = 27038.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.44029644 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0093s; samplesPerSecond = 26847.1
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.40029475 * 250; EvalErrorPrediction = 0.13200000 * 250; time = 0.0092s; samplesPerSecond = 27059.2
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.34001918 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0093s; samplesPerSecond = 26957.1
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.31615756 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0096s; samplesPerSecond = 26172.5
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.27277486 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0094s; samplesPerSecond = 26635.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.24557418 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0092s; samplesPerSecond = 27185.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.21023629 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0092s; samplesPerSecond = 27218.3
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.22380673 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0092s; samplesPerSecond = 27115.0
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20455512 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0092s; samplesPerSecond = 27068.0
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.20168480 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0091s; samplesPerSecond = 27400.3
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.19212741 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0091s; samplesPerSecond = 27397.3
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.19324124 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0091s; samplesPerSecond = 27343.3
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21777418 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0094s; samplesPerSecond = 26477.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.17514209 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0093s; samplesPerSecond = 26948.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17993773 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0091s; samplesPerSecond = 27334.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13968032 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0093s; samplesPerSecond = 26989.1
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17727753 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0095s; samplesPerSecond = 26452.2
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12898624 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0095s; samplesPerSecond = 26438.2
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21880105 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0095s; samplesPerSecond = 26340.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21850111 * 250; EvalErrorPrediction = 0.10800000 * 250; time = 0.0097s; samplesPerSecond = 25805.1
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18102491 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0096s; samplesPerSecond = 26082.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16393427 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0095s; samplesPerSecond = 26235.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13832267 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0095s; samplesPerSecond = 26241.2
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16506280 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0096s; samplesPerSecond = 25995.6
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14733234 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0095s; samplesPerSecond = 26452.2
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15041138 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0095s; samplesPerSecond = 26189.0
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12665836 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0095s; samplesPerSecond = 26296.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16643186 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0095s; samplesPerSecond = 26249.5
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14422443 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0096s; samplesPerSecond = 26147.9
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13888039 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0094s; samplesPerSecond = 26474.6
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14108686 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0095s; samplesPerSecond = 26249.5
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15887684 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0097s; samplesPerSecond = 25738.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16247402 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0094s; samplesPerSecond = 26505.5
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13586729 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0096s; samplesPerSecond = 26109.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15528679 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0094s; samplesPerSecond = 26626.9
|
||||
08/16/2016 10:01:33: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.24345139 * 10000; EvalErrorPrediction = 0.09720000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=0.379525s
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.61550018 * 250; EvalClassificationError = 0.27600000 * 250; time = 0.0108s; samplesPerSecond = 23111.8
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.59409242 * 250; EvalClassificationError = 0.28800000 * 250; time = 0.0094s; samplesPerSecond = 26612.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.53884306 * 250; EvalClassificationError = 0.20400000 * 250; time = 0.0093s; samplesPerSecond = 26890.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.52450125 * 250; EvalClassificationError = 0.15200000 * 250; time = 0.0093s; samplesPerSecond = 26942.6
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.49237463 * 250; EvalClassificationError = 0.16400000 * 250; time = 0.0092s; samplesPerSecond = 27038.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.44029644 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0093s; samplesPerSecond = 26847.1
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.40029475 * 250; EvalClassificationError = 0.13200000 * 250; time = 0.0092s; samplesPerSecond = 27059.2
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.34001918 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0093s; samplesPerSecond = 26957.1
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.31615756 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0096s; samplesPerSecond = 26172.5
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.27277486 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0094s; samplesPerSecond = 26635.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.24557418 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0092s; samplesPerSecond = 27185.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.21023629 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0092s; samplesPerSecond = 27218.3
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.22380673 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0092s; samplesPerSecond = 27115.0
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20455512 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0092s; samplesPerSecond = 27068.0
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.20168480 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0091s; samplesPerSecond = 27400.3
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.19212741 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0091s; samplesPerSecond = 27397.3
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.19324124 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0091s; samplesPerSecond = 27343.3
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21777418 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0094s; samplesPerSecond = 26477.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.17514209 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0093s; samplesPerSecond = 26948.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17993773 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0091s; samplesPerSecond = 27334.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13968032 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0093s; samplesPerSecond = 26989.1
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17727753 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0095s; samplesPerSecond = 26452.2
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12898624 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0095s; samplesPerSecond = 26438.2
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21880105 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0095s; samplesPerSecond = 26340.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21850111 * 250; EvalClassificationError = 0.10800000 * 250; time = 0.0097s; samplesPerSecond = 25805.1
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18102491 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0096s; samplesPerSecond = 26082.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16393427 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0095s; samplesPerSecond = 26235.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13832267 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0095s; samplesPerSecond = 26241.2
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16506280 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0096s; samplesPerSecond = 25995.6
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14733234 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0095s; samplesPerSecond = 26452.2
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15041138 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0095s; samplesPerSecond = 26189.0
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12665836 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0095s; samplesPerSecond = 26296.4
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16643186 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0095s; samplesPerSecond = 26249.5
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14422443 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0096s; samplesPerSecond = 26147.9
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13888039 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0094s; samplesPerSecond = 26474.6
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14108686 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0095s; samplesPerSecond = 26249.5
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15887684 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0097s; samplesPerSecond = 25738.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16247402 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0094s; samplesPerSecond = 26505.5
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13586729 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0096s; samplesPerSecond = 26109.7
|
||||
08/16/2016 10:01:33: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15528679 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0094s; samplesPerSecond = 26626.9
|
||||
08/16/2016 10:01:33: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.24345139 * 10000; EvalClassificationError = 0.09720000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=0.379525s
|
||||
08/16/2016 10:01:33: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100054.995555/Examples/Other/Simple2d_MultiGpu@release_gpu/Models/multigpu.dnn.2'
|
||||
|
||||
08/16/2016 10:01:33: Starting Epoch 3: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 2: frames [20000..30000] (first sequence at sample 20000), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:01:33: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1).
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18398525 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0097s; samplesPerSecond = 25685.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12825686 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0095s; samplesPerSecond = 26374.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17547006 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0095s; samplesPerSecond = 26318.6
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14044644 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0095s; samplesPerSecond = 26321.3
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16673170 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0094s; samplesPerSecond = 26615.6
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19317383 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0095s; samplesPerSecond = 26202.7
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12349199 * 250; EvalErrorPrediction = 0.04000000 * 250; time = 0.0093s; samplesPerSecond = 26778.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16427535 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0095s; samplesPerSecond = 26346.3
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12350212 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0093s; samplesPerSecond = 26746.5
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19958846 * 250; EvalErrorPrediction = 0.10800000 * 250; time = 0.0096s; samplesPerSecond = 26028.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14269741 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0095s; samplesPerSecond = 26189.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12369058 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0095s; samplesPerSecond = 26219.2
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16638059 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0093s; samplesPerSecond = 26847.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20047975 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0098s; samplesPerSecond = 25401.3
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.16963457 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0099s; samplesPerSecond = 25204.2
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13367401 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0098s; samplesPerSecond = 25518.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14477143 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0097s; samplesPerSecond = 25805.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21046366 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0097s; samplesPerSecond = 25791.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19247125 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0096s; samplesPerSecond = 26047.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.15027023 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0097s; samplesPerSecond = 25670.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15612870 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0098s; samplesPerSecond = 25528.4
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13684548 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0097s; samplesPerSecond = 25725.5
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17217344 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0096s; samplesPerSecond = 25939.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14419519 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0097s; samplesPerSecond = 25807.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13803181 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0097s; samplesPerSecond = 25866.5
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14209585 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0097s; samplesPerSecond = 25730.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16967141 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0097s; samplesPerSecond = 25730.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18647515 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0097s; samplesPerSecond = 25813.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16511327 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0098s; samplesPerSecond = 25541.5
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15550174 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0097s; samplesPerSecond = 25752.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18759246 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0098s; samplesPerSecond = 25525.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13178152 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0097s; samplesPerSecond = 25677.9
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14624311 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0098s; samplesPerSecond = 25583.3
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13930281 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0098s; samplesPerSecond = 25575.4
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20110083 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0099s; samplesPerSecond = 25319.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12558937 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0099s; samplesPerSecond = 25378.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18612014 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0097s; samplesPerSecond = 25821.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15336297 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0096s; samplesPerSecond = 25998.3
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11885079 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0097s; samplesPerSecond = 25850.5
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.12974982 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0096s; samplesPerSecond = 25979.4
|
||||
08/16/2016 10:01:33: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15798453 * 10000; EvalErrorPrediction = 0.07300000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=0.388464s
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18398525 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0097s; samplesPerSecond = 25685.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12825686 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0095s; samplesPerSecond = 26374.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17547006 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0095s; samplesPerSecond = 26318.6
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14044644 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0095s; samplesPerSecond = 26321.3
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16673170 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0094s; samplesPerSecond = 26615.6
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19317383 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0095s; samplesPerSecond = 26202.7
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12349199 * 250; EvalClassificationError = 0.04000000 * 250; time = 0.0093s; samplesPerSecond = 26778.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16427535 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0095s; samplesPerSecond = 26346.3
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12350212 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0093s; samplesPerSecond = 26746.5
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19958846 * 250; EvalClassificationError = 0.10800000 * 250; time = 0.0096s; samplesPerSecond = 26028.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14269741 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0095s; samplesPerSecond = 26189.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12369058 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0095s; samplesPerSecond = 26219.2
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16638059 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0093s; samplesPerSecond = 26847.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20047975 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0098s; samplesPerSecond = 25401.3
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.16963457 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0099s; samplesPerSecond = 25204.2
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13367401 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0098s; samplesPerSecond = 25518.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14477143 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0097s; samplesPerSecond = 25805.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21046366 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0097s; samplesPerSecond = 25791.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19247125 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0096s; samplesPerSecond = 26047.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.15027023 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0097s; samplesPerSecond = 25670.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15612870 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0098s; samplesPerSecond = 25528.4
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13684548 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0097s; samplesPerSecond = 25725.5
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17217344 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0096s; samplesPerSecond = 25939.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14419519 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0097s; samplesPerSecond = 25807.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13803181 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0097s; samplesPerSecond = 25866.5
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14209585 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0097s; samplesPerSecond = 25730.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16967141 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0097s; samplesPerSecond = 25730.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18647515 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0097s; samplesPerSecond = 25813.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16511327 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0098s; samplesPerSecond = 25541.5
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15550174 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0097s; samplesPerSecond = 25752.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18759246 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0098s; samplesPerSecond = 25525.8
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13178152 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0097s; samplesPerSecond = 25677.9
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14624311 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0098s; samplesPerSecond = 25583.3
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13930281 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0098s; samplesPerSecond = 25575.4
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20110083 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0099s; samplesPerSecond = 25319.0
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12558937 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0099s; samplesPerSecond = 25378.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18612014 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0097s; samplesPerSecond = 25821.1
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15336297 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0096s; samplesPerSecond = 25998.3
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11885079 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0097s; samplesPerSecond = 25850.5
|
||||
08/16/2016 10:01:33: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.12974982 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0096s; samplesPerSecond = 25979.4
|
||||
08/16/2016 10:01:33: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15798453 * 10000; EvalClassificationError = 0.07300000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=0.388464s
|
||||
08/16/2016 10:01:33: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100054.995555/Examples/Other/Simple2d_MultiGpu@release_gpu/Models/multigpu.dnn'
|
||||
08/16/2016 10:01:33: CNTKCommandTrainEnd: Multigpu_Demo_Train
|
||||
|
||||
|
|
@ -652,7 +652,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -681,7 +681,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *1] -> [2 x 1 x *1]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *1], [2 x 1] -> [2 x 1 x *1]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *1] -> [2 x 1 x *1]
|
||||
Validating --> Prior = Mean (labels) : [2 x *1] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -710,8 +710,8 @@ Memory Sharing: Out of 25 matrices, 2 are shared as 1, and 23 are not shared.
|
|||
|
||||
BlockRandomizer::StartEpoch: epoch 0: frames [0..603] (first sequence at sample 0), data subset 0 of 1
|
||||
Actual gradient aggregation time: 0.000128
|
||||
08/16/2016 10:01:33: Minibatch[1-1]: EvalErrorPrediction = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10723887 * 603
|
||||
08/16/2016 10:01:33: Final Results: Minibatch[1-1]: EvalErrorPrediction = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10723887 * 603; perplexity = 1.11320013
|
||||
08/16/2016 10:01:33: Minibatch[1-1]: EvalClassificationError = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10723887 * 603
|
||||
08/16/2016 10:01:33: Final Results: Minibatch[1-1]: EvalClassificationError = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10723887 * 603; perplexity = 1.11320013
|
||||
|
||||
08/16/2016 10:01:33: Action "test" complete.
|
||||
|
||||
|
|
|
|||
|
|
@ -80,7 +80,7 @@ Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -181,7 +181,7 @@ Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -314,7 +314,7 @@ configparameters: Multigpu.cntk:Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -394,7 +394,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -423,7 +423,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *] -> [2 x 1 x *]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *], [2 x 1] -> [2 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *] -> [2 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [2 x *] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -446,7 +446,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:19:47: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:19:47: Evaluation criterion node(s):
|
||||
08/16/2016 03:19:47: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 03:19:47: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -498,141 +498,141 @@ BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sampl
|
|||
BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sample 0), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:19:47: Starting minibatch loop.
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70264496 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0285s; samplesPerSecond = 8786.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76483063 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0246s; samplesPerSecond = 10182.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72648584 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0240s; samplesPerSecond = 10421.9
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73860254 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0238s; samplesPerSecond = 10525.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70622803 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0238s; samplesPerSecond = 10488.3
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74772852 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0242s; samplesPerSecond = 10327.6
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75092773 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0238s; samplesPerSecond = 10486.1
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78004932 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0240s; samplesPerSecond = 10434.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70444336 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0241s; samplesPerSecond = 10391.1
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69544189 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0240s; samplesPerSecond = 10398.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70595947 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0242s; samplesPerSecond = 10316.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74544189 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0234s; samplesPerSecond = 10662.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70809961 * 250; EvalErrorPrediction = 0.47600000 * 250; time = 0.0241s; samplesPerSecond = 10364.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69884375 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0241s; samplesPerSecond = 10356.3
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70363086 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0239s; samplesPerSecond = 10441.9
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69351758 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0239s; samplesPerSecond = 10447.6
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74453613 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0244s; samplesPerSecond = 10240.9
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73761426 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0242s; samplesPerSecond = 10330.6
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71868652 * 250; EvalErrorPrediction = 0.48800000 * 250; time = 0.0240s; samplesPerSecond = 10417.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71496484 * 250; EvalErrorPrediction = 0.50400000 * 250; time = 0.0236s; samplesPerSecond = 10595.0
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69961230 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0237s; samplesPerSecond = 10566.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69760645 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0238s; samplesPerSecond = 10503.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70748047 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0237s; samplesPerSecond = 10531.6
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0236s; samplesPerSecond = 10608.1
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69483203 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0246s; samplesPerSecond = 10162.6
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69258203 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0232s; samplesPerSecond = 10776.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70665625 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0239s; samplesPerSecond = 10480.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69031445 * 250; EvalErrorPrediction = 0.46800000 * 250; time = 0.0238s; samplesPerSecond = 10502.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70169531 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0240s; samplesPerSecond = 10434.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71008398 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0239s; samplesPerSecond = 10462.0
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69152930 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0238s; samplesPerSecond = 10514.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69522656 * 250; EvalErrorPrediction = 0.51600000 * 250; time = 0.0240s; samplesPerSecond = 10419.7
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69347070 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0238s; samplesPerSecond = 10490.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.68888281 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0238s; samplesPerSecond = 10499.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.68067578 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0237s; samplesPerSecond = 10557.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.66932227 * 250; EvalErrorPrediction = 0.44400000 * 250; time = 0.0242s; samplesPerSecond = 10314.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.65398437 * 250; EvalErrorPrediction = 0.24800000 * 250; time = 0.0235s; samplesPerSecond = 10638.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.63662500 * 250; EvalErrorPrediction = 0.32400000 * 250; time = 0.0234s; samplesPerSecond = 10692.4
|
||||
08/16/2016 03:19:48: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.59652344 * 250; EvalErrorPrediction = 0.04800000 * 250; time = 0.0236s; samplesPerSecond = 10595.5
|
||||
08/16/2016 03:19:48: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.55820898 * 250; EvalErrorPrediction = 0.12000000 * 250; time = 0.0245s; samplesPerSecond = 10215.8
|
||||
08/16/2016 03:19:48: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70129624 * 10000; EvalErrorPrediction = 0.46850000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.964546s
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70264496 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0285s; samplesPerSecond = 8786.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76483063 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0246s; samplesPerSecond = 10182.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72648584 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0240s; samplesPerSecond = 10421.9
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73860254 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0238s; samplesPerSecond = 10525.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70622803 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0238s; samplesPerSecond = 10488.3
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74772852 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0242s; samplesPerSecond = 10327.6
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75092773 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0238s; samplesPerSecond = 10486.1
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78004932 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0240s; samplesPerSecond = 10434.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70444336 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0241s; samplesPerSecond = 10391.1
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69544189 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0240s; samplesPerSecond = 10398.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70595947 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0242s; samplesPerSecond = 10316.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74544189 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0234s; samplesPerSecond = 10662.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70809961 * 250; EvalClassificationError = 0.47600000 * 250; time = 0.0241s; samplesPerSecond = 10364.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69884375 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0241s; samplesPerSecond = 10356.3
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70363086 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0239s; samplesPerSecond = 10441.9
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69351758 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0239s; samplesPerSecond = 10447.6
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74453613 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0244s; samplesPerSecond = 10240.9
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73761426 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0242s; samplesPerSecond = 10330.6
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71868652 * 250; EvalClassificationError = 0.48800000 * 250; time = 0.0240s; samplesPerSecond = 10417.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71496484 * 250; EvalClassificationError = 0.50400000 * 250; time = 0.0236s; samplesPerSecond = 10595.0
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69961230 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0237s; samplesPerSecond = 10566.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69760645 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0238s; samplesPerSecond = 10503.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70748047 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0237s; samplesPerSecond = 10531.6
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0236s; samplesPerSecond = 10608.1
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69483203 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0246s; samplesPerSecond = 10162.6
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69258203 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0232s; samplesPerSecond = 10776.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70665625 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0239s; samplesPerSecond = 10480.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69031445 * 250; EvalClassificationError = 0.46800000 * 250; time = 0.0238s; samplesPerSecond = 10502.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70169531 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0240s; samplesPerSecond = 10434.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71008398 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0239s; samplesPerSecond = 10462.0
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69152930 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0238s; samplesPerSecond = 10514.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69522656 * 250; EvalClassificationError = 0.51600000 * 250; time = 0.0240s; samplesPerSecond = 10419.7
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69347070 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0238s; samplesPerSecond = 10490.5
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.68888281 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0238s; samplesPerSecond = 10499.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.68067578 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0237s; samplesPerSecond = 10557.4
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.66932227 * 250; EvalClassificationError = 0.44400000 * 250; time = 0.0242s; samplesPerSecond = 10314.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.65398437 * 250; EvalClassificationError = 0.24800000 * 250; time = 0.0235s; samplesPerSecond = 10638.8
|
||||
08/16/2016 03:19:47: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.63662500 * 250; EvalClassificationError = 0.32400000 * 250; time = 0.0234s; samplesPerSecond = 10692.4
|
||||
08/16/2016 03:19:48: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.59652344 * 250; EvalClassificationError = 0.04800000 * 250; time = 0.0236s; samplesPerSecond = 10595.5
|
||||
08/16/2016 03:19:48: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.55820898 * 250; EvalClassificationError = 0.12000000 * 250; time = 0.0245s; samplesPerSecond = 10215.8
|
||||
08/16/2016 03:19:48: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70129624 * 10000; EvalClassificationError = 0.46850000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.964546s
|
||||
08/16/2016 03:19:48: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031849.416502\Examples\Other\Simple2d_MultiGpu@release_cpu/Models/multigpu.dnn.1'
|
||||
|
||||
08/16/2016 03:19:48: Starting Epoch 2: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 1: frames [10000..20000] (first sequence at sample 10000), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:19:48: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1).
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.50509082 * 250; EvalErrorPrediction = 0.14400000 * 250; time = 0.0250s; samplesPerSecond = 9991.2
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.45891377 * 250; EvalErrorPrediction = 0.13200000 * 250; time = 0.0251s; samplesPerSecond = 9958.6
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.38371187 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0247s; samplesPerSecond = 10117.4
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.35526704 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0254s; samplesPerSecond = 9837.5
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.31361566 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0249s; samplesPerSecond = 10049.0
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.29756372 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0254s; samplesPerSecond = 9831.3
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.27214716 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0245s; samplesPerSecond = 10219.1
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.23149490 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0244s; samplesPerSecond = 10231.2
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.23825536 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0247s; samplesPerSecond = 10102.6
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.21847410 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0251s; samplesPerSecond = 9945.5
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.19974600 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0248s; samplesPerSecond = 10088.4
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.18213383 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0252s; samplesPerSecond = 9934.0
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.19621664 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0250s; samplesPerSecond = 10018.4
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.18917135 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0266s; samplesPerSecond = 9390.4
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.18997701 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0272s; samplesPerSecond = 9179.0
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.18456273 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0293s; samplesPerSecond = 8534.2
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.18678577 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0274s; samplesPerSecond = 9126.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21314113 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0270s; samplesPerSecond = 9242.5
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.16860178 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0281s; samplesPerSecond = 8903.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17451651 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0292s; samplesPerSecond = 8561.1
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13649532 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0291s; samplesPerSecond = 8585.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17557703 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0271s; samplesPerSecond = 9213.2
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12777527 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0266s; samplesPerSecond = 9414.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21833707 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0245s; samplesPerSecond = 10188.7
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21788590 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0251s; samplesPerSecond = 9969.7
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18130830 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0250s; samplesPerSecond = 9987.6
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16267770 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0249s; samplesPerSecond = 10056.7
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13704118 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0247s; samplesPerSecond = 10125.1
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16545012 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0242s; samplesPerSecond = 10321.6
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14842740 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0252s; samplesPerSecond = 9932.1
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15099778 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0250s; samplesPerSecond = 9988.0
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12730237 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0256s; samplesPerSecond = 9775.2
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16464377 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0257s; samplesPerSecond = 9723.5
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14324668 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0248s; samplesPerSecond = 10096.5
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13824633 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0254s; samplesPerSecond = 9853.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14128747 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0248s; samplesPerSecond = 10079.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15910150 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0253s; samplesPerSecond = 9863.1
|
||||
08/16/2016 03:19:49: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16253611 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0251s; samplesPerSecond = 9950.6
|
||||
08/16/2016 03:19:49: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13535163 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0256s; samplesPerSecond = 9772.1
|
||||
08/16/2016 03:19:49: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15552570 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0249s; samplesPerSecond = 10044.2
|
||||
08/16/2016 03:19:49: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.20771504 * 10000; EvalErrorPrediction = 0.08060000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=1.02956s
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.50509082 * 250; EvalClassificationError = 0.14400000 * 250; time = 0.0250s; samplesPerSecond = 9991.2
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.45891377 * 250; EvalClassificationError = 0.13200000 * 250; time = 0.0251s; samplesPerSecond = 9958.6
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.38371187 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0247s; samplesPerSecond = 10117.4
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.35526704 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0254s; samplesPerSecond = 9837.5
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.31361566 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0249s; samplesPerSecond = 10049.0
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.29756372 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0254s; samplesPerSecond = 9831.3
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.27214716 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0245s; samplesPerSecond = 10219.1
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.23149490 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0244s; samplesPerSecond = 10231.2
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.23825536 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0247s; samplesPerSecond = 10102.6
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.21847410 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0251s; samplesPerSecond = 9945.5
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.19974600 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0248s; samplesPerSecond = 10088.4
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.18213383 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0252s; samplesPerSecond = 9934.0
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.19621664 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0250s; samplesPerSecond = 10018.4
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.18917135 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0266s; samplesPerSecond = 9390.4
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.18997701 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0272s; samplesPerSecond = 9179.0
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.18456273 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0293s; samplesPerSecond = 8534.2
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.18678577 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0274s; samplesPerSecond = 9126.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21314113 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0270s; samplesPerSecond = 9242.5
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.16860178 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0281s; samplesPerSecond = 8903.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17451651 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0292s; samplesPerSecond = 8561.1
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13649532 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0291s; samplesPerSecond = 8585.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17557703 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0271s; samplesPerSecond = 9213.2
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12777527 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0266s; samplesPerSecond = 9414.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21833707 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0245s; samplesPerSecond = 10188.7
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21788590 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0251s; samplesPerSecond = 9969.7
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18130830 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0250s; samplesPerSecond = 9987.6
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16267770 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0249s; samplesPerSecond = 10056.7
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13704118 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0247s; samplesPerSecond = 10125.1
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16545012 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0242s; samplesPerSecond = 10321.6
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14842740 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0252s; samplesPerSecond = 9932.1
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15099778 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0250s; samplesPerSecond = 9988.0
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12730237 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0256s; samplesPerSecond = 9775.2
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16464377 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0257s; samplesPerSecond = 9723.5
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14324668 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0248s; samplesPerSecond = 10096.5
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13824633 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0254s; samplesPerSecond = 9853.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14128747 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0248s; samplesPerSecond = 10079.8
|
||||
08/16/2016 03:19:48: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15910150 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0253s; samplesPerSecond = 9863.1
|
||||
08/16/2016 03:19:49: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16253611 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0251s; samplesPerSecond = 9950.6
|
||||
08/16/2016 03:19:49: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13535163 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0256s; samplesPerSecond = 9772.1
|
||||
08/16/2016 03:19:49: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15552570 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0249s; samplesPerSecond = 10044.2
|
||||
08/16/2016 03:19:49: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.20771504 * 10000; EvalClassificationError = 0.08060000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=1.02956s
|
||||
08/16/2016 03:19:49: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031849.416502\Examples\Other\Simple2d_MultiGpu@release_cpu/Models/multigpu.dnn.2'
|
||||
|
||||
08/16/2016 03:19:49: Starting Epoch 3: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 2: frames [20000..30000] (first sequence at sample 20000), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:19:49: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1).
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18436522 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0246s; samplesPerSecond = 10145.7
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12821186 * 250; EvalErrorPrediction = 0.04800000 * 250; time = 0.0251s; samplesPerSecond = 9945.1
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17512306 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0248s; samplesPerSecond = 10084.3
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.13980331 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0246s; samplesPerSecond = 10172.5
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16538291 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0247s; samplesPerSecond = 10124.3
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19375913 * 250; EvalErrorPrediction = 0.11600000 * 250; time = 0.0256s; samplesPerSecond = 9764.1
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12331922 * 250; EvalErrorPrediction = 0.04000000 * 250; time = 0.0254s; samplesPerSecond = 9851.8
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16604588 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0258s; samplesPerSecond = 9702.7
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12468993 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0249s; samplesPerSecond = 10048.6
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.20005103 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0253s; samplesPerSecond = 9889.2
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14282824 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0242s; samplesPerSecond = 10340.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12364929 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0243s; samplesPerSecond = 10295.7
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16738214 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0252s; samplesPerSecond = 9906.5
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19934515 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0266s; samplesPerSecond = 9392.5
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.16932168 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0247s; samplesPerSecond = 10128.4
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13332017 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0247s; samplesPerSecond = 10125.6
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14351372 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0248s; samplesPerSecond = 10100.6
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20938709 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0247s; samplesPerSecond = 10107.5
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19203984 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0252s; samplesPerSecond = 9921.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.15014813 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0250s; samplesPerSecond = 10010.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15581546 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0249s; samplesPerSecond = 10054.3
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13716517 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0249s; samplesPerSecond = 10047.8
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17233280 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0262s; samplesPerSecond = 9559.1
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14434328 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0253s; samplesPerSecond = 9878.3
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13849430 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0246s; samplesPerSecond = 10182.9
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14141637 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0242s; samplesPerSecond = 10331.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16967658 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0252s; samplesPerSecond = 9932.9
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18536492 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0248s; samplesPerSecond = 10077.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16547838 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0248s; samplesPerSecond = 10073.7
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15382617 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0250s; samplesPerSecond = 9985.2
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18866317 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0250s; samplesPerSecond = 9980.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13254335 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0249s; samplesPerSecond = 10049.8
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14548822 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0245s; samplesPerSecond = 10191.2
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13912198 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0245s; samplesPerSecond = 10194.1
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20068190 * 250; EvalErrorPrediction = 0.10800000 * 250; time = 0.0257s; samplesPerSecond = 9729.9
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12564777 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0245s; samplesPerSecond = 10190.8
|
||||
08/16/2016 03:19:50: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18466509 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0251s; samplesPerSecond = 9966.1
|
||||
08/16/2016 03:19:50: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15248240 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0243s; samplesPerSecond = 10290.2
|
||||
08/16/2016 03:19:50: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11889087 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0245s; samplesPerSecond = 10185.0
|
||||
08/16/2016 03:19:50: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.12990310 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0252s; samplesPerSecond = 9902.2
|
||||
08/16/2016 03:19:50: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15784221 * 10000; EvalErrorPrediction = 0.07350000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=1.00011s
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18436522 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0246s; samplesPerSecond = 10145.7
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12821186 * 250; EvalClassificationError = 0.04800000 * 250; time = 0.0251s; samplesPerSecond = 9945.1
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17512306 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0248s; samplesPerSecond = 10084.3
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.13980331 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0246s; samplesPerSecond = 10172.5
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16538291 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0247s; samplesPerSecond = 10124.3
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19375913 * 250; EvalClassificationError = 0.11600000 * 250; time = 0.0256s; samplesPerSecond = 9764.1
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12331922 * 250; EvalClassificationError = 0.04000000 * 250; time = 0.0254s; samplesPerSecond = 9851.8
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16604588 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0258s; samplesPerSecond = 9702.7
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12468993 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0249s; samplesPerSecond = 10048.6
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.20005103 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0253s; samplesPerSecond = 9889.2
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14282824 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0242s; samplesPerSecond = 10340.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12364929 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0243s; samplesPerSecond = 10295.7
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16738214 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0252s; samplesPerSecond = 9906.5
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19934515 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0266s; samplesPerSecond = 9392.5
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.16932168 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0247s; samplesPerSecond = 10128.4
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13332017 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0247s; samplesPerSecond = 10125.6
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14351372 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0248s; samplesPerSecond = 10100.6
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20938709 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0247s; samplesPerSecond = 10107.5
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19203984 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0252s; samplesPerSecond = 9921.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.15014813 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0250s; samplesPerSecond = 10010.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15581546 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0249s; samplesPerSecond = 10054.3
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13716517 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0249s; samplesPerSecond = 10047.8
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17233280 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0262s; samplesPerSecond = 9559.1
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14434328 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0253s; samplesPerSecond = 9878.3
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13849430 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0246s; samplesPerSecond = 10182.9
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14141637 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0242s; samplesPerSecond = 10331.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16967658 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0252s; samplesPerSecond = 9932.9
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18536492 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0248s; samplesPerSecond = 10077.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16547838 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0248s; samplesPerSecond = 10073.7
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15382617 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0250s; samplesPerSecond = 9985.2
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18866317 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0250s; samplesPerSecond = 9980.0
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13254335 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0249s; samplesPerSecond = 10049.8
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14548822 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0245s; samplesPerSecond = 10191.2
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13912198 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0245s; samplesPerSecond = 10194.1
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20068190 * 250; EvalClassificationError = 0.10800000 * 250; time = 0.0257s; samplesPerSecond = 9729.9
|
||||
08/16/2016 03:19:49: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12564777 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0245s; samplesPerSecond = 10190.8
|
||||
08/16/2016 03:19:50: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18466509 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0251s; samplesPerSecond = 9966.1
|
||||
08/16/2016 03:19:50: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15248240 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0243s; samplesPerSecond = 10290.2
|
||||
08/16/2016 03:19:50: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11889087 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0245s; samplesPerSecond = 10185.0
|
||||
08/16/2016 03:19:50: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.12990310 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0252s; samplesPerSecond = 9902.2
|
||||
08/16/2016 03:19:50: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15784221 * 10000; EvalClassificationError = 0.07350000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=1.00011s
|
||||
08/16/2016 03:19:50: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031849.416502\Examples\Other\Simple2d_MultiGpu@release_cpu/Models/multigpu.dnn'
|
||||
08/16/2016 03:19:50: CNTKCommandTrainEnd: Multigpu_Demo_Train
|
||||
|
||||
|
|
@ -650,7 +650,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -679,7 +679,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *1] -> [2 x 1 x *1]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *1], [2 x 1] -> [2 x 1 x *1]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *1] -> [2 x 1 x *1]
|
||||
Validating --> Prior = Mean (labels) : [2 x *1] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -708,8 +708,8 @@ Memory Sharing: Out of 25 matrices, 2 are shared as 1, and 23 are not shared.
|
|||
|
||||
BlockRandomizer::StartEpoch: epoch 0: frames [0..603] (first sequence at sample 0), data subset 0 of 1
|
||||
Actual gradient aggregation time: 0.000128
|
||||
08/16/2016 03:19:50: Minibatch[1-1]: EvalErrorPrediction = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10640968 * 603
|
||||
08/16/2016 03:19:50: Final Results: Minibatch[1-1]: EvalErrorPrediction = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10640968 * 603; perplexity = 1.11227746
|
||||
08/16/2016 03:19:50: Minibatch[1-1]: EvalClassificationError = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10640968 * 603
|
||||
08/16/2016 03:19:50: Final Results: Minibatch[1-1]: EvalClassificationError = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10640968 * 603; perplexity = 1.11227746
|
||||
|
||||
08/16/2016 03:19:50: Action "test" complete.
|
||||
|
||||
|
|
|
|||
|
|
@ -80,7 +80,7 @@ Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -181,7 +181,7 @@ Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -314,7 +314,7 @@ configparameters: Multigpu.cntk:Multigpu_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -395,7 +395,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -424,7 +424,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *] -> [2 x 1 x *]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *], [2 x 1] -> [2 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *] -> [2 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [2 x *] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -447,7 +447,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:19:54: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:19:54: Evaluation criterion node(s):
|
||||
08/16/2016 03:19:54: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 03:19:54: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -499,141 +499,141 @@ BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sampl
|
|||
BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sample 0), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:19:54: Starting minibatch loop.
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70124231 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0171s; samplesPerSecond = 14629.3
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76372424 * 250; EvalErrorPrediction = 0.46400000 * 250; time = 0.0156s; samplesPerSecond = 15976.5
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72703027 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0158s; samplesPerSecond = 15853.9
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73895923 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0157s; samplesPerSecond = 15952.0
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70621924 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0157s; samplesPerSecond = 15907.4
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74767041 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0158s; samplesPerSecond = 15831.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75094434 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0158s; samplesPerSecond = 15822.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78058936 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0157s; samplesPerSecond = 15880.1
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70407129 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0157s; samplesPerSecond = 15927.6
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69555762 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0157s; samplesPerSecond = 15926.6
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70626123 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0158s; samplesPerSecond = 15816.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74540430 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0157s; samplesPerSecond = 15884.1
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70824414 * 250; EvalErrorPrediction = 0.47600000 * 250; time = 0.0158s; samplesPerSecond = 15815.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69895020 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0157s; samplesPerSecond = 15895.2
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70353223 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0157s; samplesPerSecond = 15937.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69346387 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0158s; samplesPerSecond = 15825.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74449902 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0157s; samplesPerSecond = 15903.3
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73767969 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0157s; samplesPerSecond = 15895.2
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71876855 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0157s; samplesPerSecond = 15889.2
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71509473 * 250; EvalErrorPrediction = 0.50400000 * 250; time = 0.0158s; samplesPerSecond = 15836.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69956152 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0157s; samplesPerSecond = 15888.1
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0157s; samplesPerSecond = 15917.5
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70736035 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0157s; samplesPerSecond = 15923.6
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69820508 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0158s; samplesPerSecond = 15839.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69537109 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0156s; samplesPerSecond = 15981.6
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69347266 * 250; EvalErrorPrediction = 0.51200000 * 250; time = 0.0162s; samplesPerSecond = 15477.0
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70801172 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0157s; samplesPerSecond = 15921.5
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69131641 * 250; EvalErrorPrediction = 0.46800000 * 250; time = 0.0158s; samplesPerSecond = 15823.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70370312 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0157s; samplesPerSecond = 15923.6
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71200195 * 250; EvalErrorPrediction = 0.43600000 * 250; time = 0.0157s; samplesPerSecond = 15900.3
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69506836 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0158s; samplesPerSecond = 15838.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69935352 * 250; EvalErrorPrediction = 0.51600000 * 250; time = 0.0158s; samplesPerSecond = 15830.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69887109 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0158s; samplesPerSecond = 15833.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69604492 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0156s; samplesPerSecond = 15991.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.69011719 * 250; EvalErrorPrediction = 0.48800000 * 250; time = 0.0158s; samplesPerSecond = 15794.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.68419531 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0158s; samplesPerSecond = 15850.9
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.67551367 * 250; EvalErrorPrediction = 0.32400000 * 250; time = 0.0158s; samplesPerSecond = 15859.9
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.67028516 * 250; EvalErrorPrediction = 0.40000000 * 250; time = 0.0157s; samplesPerSecond = 15940.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.65152734 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0158s; samplesPerSecond = 15841.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.63594727 * 250; EvalErrorPrediction = 0.22000000 * 250; time = 0.0157s; samplesPerSecond = 15917.5
|
||||
08/16/2016 03:19:55: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70729233 * 10000; EvalErrorPrediction = 0.47740000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.633908s
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70124231 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0171s; samplesPerSecond = 14629.3
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76372424 * 250; EvalClassificationError = 0.46400000 * 250; time = 0.0156s; samplesPerSecond = 15976.5
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72703027 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0158s; samplesPerSecond = 15853.9
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73895923 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0157s; samplesPerSecond = 15952.0
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70621924 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0157s; samplesPerSecond = 15907.4
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74767041 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0158s; samplesPerSecond = 15831.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75094434 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0158s; samplesPerSecond = 15822.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78058936 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0157s; samplesPerSecond = 15880.1
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70407129 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0157s; samplesPerSecond = 15927.6
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69555762 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0157s; samplesPerSecond = 15926.6
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70626123 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0158s; samplesPerSecond = 15816.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74540430 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0157s; samplesPerSecond = 15884.1
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70824414 * 250; EvalClassificationError = 0.47600000 * 250; time = 0.0158s; samplesPerSecond = 15815.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69895020 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0157s; samplesPerSecond = 15895.2
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70353223 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0157s; samplesPerSecond = 15937.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69346387 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0158s; samplesPerSecond = 15825.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74449902 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0157s; samplesPerSecond = 15903.3
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73767969 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0157s; samplesPerSecond = 15895.2
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71876855 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0157s; samplesPerSecond = 15889.2
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71509473 * 250; EvalClassificationError = 0.50400000 * 250; time = 0.0158s; samplesPerSecond = 15836.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69956152 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0157s; samplesPerSecond = 15888.1
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0157s; samplesPerSecond = 15917.5
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70736035 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0157s; samplesPerSecond = 15923.6
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69820508 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0158s; samplesPerSecond = 15839.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69537109 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0156s; samplesPerSecond = 15981.6
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69347266 * 250; EvalClassificationError = 0.51200000 * 250; time = 0.0162s; samplesPerSecond = 15477.0
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70801172 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0157s; samplesPerSecond = 15921.5
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69131641 * 250; EvalClassificationError = 0.46800000 * 250; time = 0.0158s; samplesPerSecond = 15823.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70370312 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0157s; samplesPerSecond = 15923.6
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71200195 * 250; EvalClassificationError = 0.43600000 * 250; time = 0.0157s; samplesPerSecond = 15900.3
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69506836 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0158s; samplesPerSecond = 15838.8
|
||||
08/16/2016 03:19:54: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69935352 * 250; EvalClassificationError = 0.51600000 * 250; time = 0.0158s; samplesPerSecond = 15830.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69887109 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0158s; samplesPerSecond = 15833.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69604492 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0156s; samplesPerSecond = 15991.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.69011719 * 250; EvalClassificationError = 0.48800000 * 250; time = 0.0158s; samplesPerSecond = 15794.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.68419531 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0158s; samplesPerSecond = 15850.9
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.67551367 * 250; EvalClassificationError = 0.32400000 * 250; time = 0.0158s; samplesPerSecond = 15859.9
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.67028516 * 250; EvalClassificationError = 0.40000000 * 250; time = 0.0157s; samplesPerSecond = 15940.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.65152734 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0158s; samplesPerSecond = 15841.8
|
||||
08/16/2016 03:19:55: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.63594727 * 250; EvalClassificationError = 0.22000000 * 250; time = 0.0157s; samplesPerSecond = 15917.5
|
||||
08/16/2016 03:19:55: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70729233 * 10000; EvalClassificationError = 0.47740000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.633908s
|
||||
08/16/2016 03:19:55: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031849.416502\Examples\Other\Simple2d_MultiGpu@release_gpu/Models/multigpu.dnn.1'
|
||||
|
||||
08/16/2016 03:19:55: Starting Epoch 2: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 1: frames [10000..20000] (first sequence at sample 10000), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:19:55: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1).
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.61550018 * 250; EvalErrorPrediction = 0.27600000 * 250; time = 0.0399s; samplesPerSecond = 6268.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.59409242 * 250; EvalErrorPrediction = 0.28800000 * 250; time = 0.0380s; samplesPerSecond = 6577.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.53884306 * 250; EvalErrorPrediction = 0.20400000 * 250; time = 0.0379s; samplesPerSecond = 6604.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.52450125 * 250; EvalErrorPrediction = 0.15200000 * 250; time = 0.0374s; samplesPerSecond = 6683.4
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.49237463 * 250; EvalErrorPrediction = 0.16400000 * 250; time = 0.0374s; samplesPerSecond = 6678.4
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.44029644 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0376s; samplesPerSecond = 6645.4
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.40029475 * 250; EvalErrorPrediction = 0.13200000 * 250; time = 0.0370s; samplesPerSecond = 6763.7
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.34001918 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0378s; samplesPerSecond = 6611.8
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.31615756 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0380s; samplesPerSecond = 6582.1
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.27277486 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0375s; samplesPerSecond = 6672.4
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.24557418 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0375s; samplesPerSecond = 6662.2
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.21023629 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0372s; samplesPerSecond = 6712.5
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.22380673 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0376s; samplesPerSecond = 6640.6
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20455512 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0375s; samplesPerSecond = 6666.1
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.20168480 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0377s; samplesPerSecond = 6623.4
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.19212741 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0373s; samplesPerSecond = 6699.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.19324124 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0374s; samplesPerSecond = 6679.5
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21777418 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0372s; samplesPerSecond = 6729.3
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.17514209 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0376s; samplesPerSecond = 6644.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17993773 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0376s; samplesPerSecond = 6649.8
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13968032 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0376s; samplesPerSecond = 6641.3
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17727753 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0373s; samplesPerSecond = 6699.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12898624 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0370s; samplesPerSecond = 6749.8
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21880105 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0373s; samplesPerSecond = 6708.2
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21850111 * 250; EvalErrorPrediction = 0.10800000 * 250; time = 0.0377s; samplesPerSecond = 6622.9
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18102491 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0377s; samplesPerSecond = 6636.6
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16393427 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0372s; samplesPerSecond = 6714.3
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13832267 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0375s; samplesPerSecond = 6659.7
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16506280 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0377s; samplesPerSecond = 6634.6
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14733234 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0376s; samplesPerSecond = 6644.7
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15041138 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0379s; samplesPerSecond = 6600.5
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12665836 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0376s; samplesPerSecond = 6641.3
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16643186 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0373s; samplesPerSecond = 6699.5
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14422443 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0360s; samplesPerSecond = 6946.8
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13888039 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0364s; samplesPerSecond = 6860.0
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14108686 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0377s; samplesPerSecond = 6629.0
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15887684 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0375s; samplesPerSecond = 6662.6
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16247402 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0377s; samplesPerSecond = 6630.8
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13586729 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0377s; samplesPerSecond = 6631.1
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15528679 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0376s; samplesPerSecond = 6642.8
|
||||
08/16/2016 03:19:56: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.24345139 * 10000; EvalErrorPrediction = 0.09720000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=1.50329s
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.61550018 * 250; EvalClassificationError = 0.27600000 * 250; time = 0.0399s; samplesPerSecond = 6268.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.59409242 * 250; EvalClassificationError = 0.28800000 * 250; time = 0.0380s; samplesPerSecond = 6577.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.53884306 * 250; EvalClassificationError = 0.20400000 * 250; time = 0.0379s; samplesPerSecond = 6604.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.52450125 * 250; EvalClassificationError = 0.15200000 * 250; time = 0.0374s; samplesPerSecond = 6683.4
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.49237463 * 250; EvalClassificationError = 0.16400000 * 250; time = 0.0374s; samplesPerSecond = 6678.4
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.44029644 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0376s; samplesPerSecond = 6645.4
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.40029475 * 250; EvalClassificationError = 0.13200000 * 250; time = 0.0370s; samplesPerSecond = 6763.7
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.34001918 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0378s; samplesPerSecond = 6611.8
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.31615756 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0380s; samplesPerSecond = 6582.1
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.27277486 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0375s; samplesPerSecond = 6672.4
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.24557418 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0375s; samplesPerSecond = 6662.2
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.21023629 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0372s; samplesPerSecond = 6712.5
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.22380673 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0376s; samplesPerSecond = 6640.6
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20455512 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0375s; samplesPerSecond = 6666.1
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.20168480 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0377s; samplesPerSecond = 6623.4
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.19212741 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0373s; samplesPerSecond = 6699.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.19324124 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0374s; samplesPerSecond = 6679.5
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21777418 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0372s; samplesPerSecond = 6729.3
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.17514209 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0376s; samplesPerSecond = 6644.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17993773 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0376s; samplesPerSecond = 6649.8
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13968032 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0376s; samplesPerSecond = 6641.3
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17727753 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0373s; samplesPerSecond = 6699.0
|
||||
08/16/2016 03:19:55: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12898624 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0370s; samplesPerSecond = 6749.8
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21880105 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0373s; samplesPerSecond = 6708.2
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21850111 * 250; EvalClassificationError = 0.10800000 * 250; time = 0.0377s; samplesPerSecond = 6622.9
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18102491 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0377s; samplesPerSecond = 6636.6
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16393427 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0372s; samplesPerSecond = 6714.3
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13832267 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0375s; samplesPerSecond = 6659.7
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16506280 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0377s; samplesPerSecond = 6634.6
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14733234 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0376s; samplesPerSecond = 6644.7
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15041138 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0379s; samplesPerSecond = 6600.5
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12665836 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0376s; samplesPerSecond = 6641.3
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16643186 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0373s; samplesPerSecond = 6699.5
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14422443 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0360s; samplesPerSecond = 6946.8
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13888039 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0364s; samplesPerSecond = 6860.0
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14108686 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0377s; samplesPerSecond = 6629.0
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15887684 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0375s; samplesPerSecond = 6662.6
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16247402 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0377s; samplesPerSecond = 6630.8
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13586729 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0377s; samplesPerSecond = 6631.1
|
||||
08/16/2016 03:19:56: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15528679 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0376s; samplesPerSecond = 6642.8
|
||||
08/16/2016 03:19:56: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.24345139 * 10000; EvalClassificationError = 0.09720000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=1.50329s
|
||||
08/16/2016 03:19:56: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031849.416502\Examples\Other\Simple2d_MultiGpu@release_gpu/Models/multigpu.dnn.2'
|
||||
|
||||
08/16/2016 03:19:56: Starting Epoch 3: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 2: frames [20000..30000] (first sequence at sample 20000), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:19:56: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1).
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18398525 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0376s; samplesPerSecond = 6641.3
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12825686 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0376s; samplesPerSecond = 6653.0
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17547006 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0374s; samplesPerSecond = 6692.7
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14044644 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0373s; samplesPerSecond = 6703.9
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16673170 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0379s; samplesPerSecond = 6591.3
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19317383 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0374s; samplesPerSecond = 6678.2
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12349199 * 250; EvalErrorPrediction = 0.04000000 * 250; time = 0.0381s; samplesPerSecond = 6555.0
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16427535 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0373s; samplesPerSecond = 6693.6
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12350212 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0376s; samplesPerSecond = 6652.3
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19958846 * 250; EvalErrorPrediction = 0.10800000 * 250; time = 0.0377s; samplesPerSecond = 6625.1
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14269741 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0378s; samplesPerSecond = 6615.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12369058 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0370s; samplesPerSecond = 6755.8
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16638059 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0375s; samplesPerSecond = 6669.5
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20047975 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0373s; samplesPerSecond = 6704.2
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.16963457 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0371s; samplesPerSecond = 6744.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13367401 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0372s; samplesPerSecond = 6724.0
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14477143 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0369s; samplesPerSecond = 6775.3
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21046366 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0373s; samplesPerSecond = 6702.8
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19247125 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0374s; samplesPerSecond = 6679.8
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.15027023 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0371s; samplesPerSecond = 6747.5
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15612870 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0377s; samplesPerSecond = 6635.9
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13684548 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0373s; samplesPerSecond = 6697.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17217344 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0377s; samplesPerSecond = 6638.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14419519 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0375s; samplesPerSecond = 6666.8
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13803181 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0375s; samplesPerSecond = 6659.6
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14209585 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0374s; samplesPerSecond = 6680.2
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16967141 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0373s; samplesPerSecond = 6710.0
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18647515 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0377s; samplesPerSecond = 6630.2
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16511327 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0363s; samplesPerSecond = 6885.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15550174 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0376s; samplesPerSecond = 6646.5
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18759246 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0373s; samplesPerSecond = 6695.4
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13178152 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0376s; samplesPerSecond = 6657.3
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14624311 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0372s; samplesPerSecond = 6714.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13930281 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0374s; samplesPerSecond = 6682.3
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20110083 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0377s; samplesPerSecond = 6632.4
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12558937 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0369s; samplesPerSecond = 6776.5
|
||||
08/16/2016 03:19:58: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18612014 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0376s; samplesPerSecond = 6647.2
|
||||
08/16/2016 03:19:58: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15336297 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0374s; samplesPerSecond = 6684.3
|
||||
08/16/2016 03:19:58: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11885079 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0375s; samplesPerSecond = 6668.4
|
||||
08/16/2016 03:19:58: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.12974982 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0381s; samplesPerSecond = 6568.7
|
||||
08/16/2016 03:19:58: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15798453 * 10000; EvalErrorPrediction = 0.07300000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=1.49905s
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18398525 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0376s; samplesPerSecond = 6641.3
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12825686 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0376s; samplesPerSecond = 6653.0
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17547006 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0374s; samplesPerSecond = 6692.7
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14044644 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0373s; samplesPerSecond = 6703.9
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16673170 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0379s; samplesPerSecond = 6591.3
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19317383 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0374s; samplesPerSecond = 6678.2
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12349199 * 250; EvalClassificationError = 0.04000000 * 250; time = 0.0381s; samplesPerSecond = 6555.0
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16427535 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0373s; samplesPerSecond = 6693.6
|
||||
08/16/2016 03:19:56: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12350212 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0376s; samplesPerSecond = 6652.3
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19958846 * 250; EvalClassificationError = 0.10800000 * 250; time = 0.0377s; samplesPerSecond = 6625.1
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14269741 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0378s; samplesPerSecond = 6615.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12369058 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0370s; samplesPerSecond = 6755.8
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16638059 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0375s; samplesPerSecond = 6669.5
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20047975 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0373s; samplesPerSecond = 6704.2
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.16963457 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0371s; samplesPerSecond = 6744.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13367401 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0372s; samplesPerSecond = 6724.0
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14477143 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0369s; samplesPerSecond = 6775.3
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21046366 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0373s; samplesPerSecond = 6702.8
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19247125 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0374s; samplesPerSecond = 6679.8
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.15027023 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0371s; samplesPerSecond = 6747.5
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15612870 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0377s; samplesPerSecond = 6635.9
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13684548 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0373s; samplesPerSecond = 6697.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17217344 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0377s; samplesPerSecond = 6638.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14419519 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0375s; samplesPerSecond = 6666.8
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13803181 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0375s; samplesPerSecond = 6659.6
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14209585 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0374s; samplesPerSecond = 6680.2
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16967141 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0373s; samplesPerSecond = 6710.0
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18647515 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0377s; samplesPerSecond = 6630.2
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16511327 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0363s; samplesPerSecond = 6885.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15550174 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0376s; samplesPerSecond = 6646.5
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18759246 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0373s; samplesPerSecond = 6695.4
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13178152 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0376s; samplesPerSecond = 6657.3
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14624311 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0372s; samplesPerSecond = 6714.7
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13930281 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0374s; samplesPerSecond = 6682.3
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20110083 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0377s; samplesPerSecond = 6632.4
|
||||
08/16/2016 03:19:57: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12558937 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0369s; samplesPerSecond = 6776.5
|
||||
08/16/2016 03:19:58: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18612014 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0376s; samplesPerSecond = 6647.2
|
||||
08/16/2016 03:19:58: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15336297 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0374s; samplesPerSecond = 6684.3
|
||||
08/16/2016 03:19:58: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11885079 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0375s; samplesPerSecond = 6668.4
|
||||
08/16/2016 03:19:58: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.12974982 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0381s; samplesPerSecond = 6568.7
|
||||
08/16/2016 03:19:58: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15798453 * 10000; EvalClassificationError = 0.07300000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=1.49905s
|
||||
08/16/2016 03:19:58: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031849.416502\Examples\Other\Simple2d_MultiGpu@release_gpu/Models/multigpu.dnn'
|
||||
08/16/2016 03:19:58: CNTKCommandTrainEnd: Multigpu_Demo_Train
|
||||
|
||||
|
|
@ -651,7 +651,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -680,7 +680,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *1] -> [2 x 1 x *1]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *1], [2 x 1] -> [2 x 1 x *1]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *1] -> [2 x 1 x *1]
|
||||
Validating --> Prior = Mean (labels) : [2 x *1] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -709,8 +709,8 @@ Memory Sharing: Out of 25 matrices, 2 are shared as 1, and 23 are not shared.
|
|||
|
||||
BlockRandomizer::StartEpoch: epoch 0: frames [0..603] (first sequence at sample 0), data subset 0 of 1
|
||||
Actual gradient aggregation time: 5.7e-005
|
||||
08/16/2016 03:19:58: Minibatch[1-1]: EvalErrorPrediction = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10723887 * 603
|
||||
08/16/2016 03:19:58: Final Results: Minibatch[1-1]: EvalErrorPrediction = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10723887 * 603; perplexity = 1.11320013
|
||||
08/16/2016 03:19:58: Minibatch[1-1]: EvalClassificationError = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10723887 * 603
|
||||
08/16/2016 03:19:58: Final Results: Minibatch[1-1]: EvalClassificationError = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10723887 * 603; perplexity = 1.11320013
|
||||
|
||||
08/16/2016 03:19:58: Action "test" complete.
|
||||
|
||||
|
|
|
|||
|
|
@ -71,7 +71,7 @@ Simple_Demo_Train = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -170,7 +170,7 @@ Simple_Demo_Train = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -313,7 +313,7 @@ configparameters: Simple.cntk:Simple_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -380,7 +380,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -409,7 +409,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *] -> [2 x 1 x *]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *], [2 x 1] -> [2 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *] -> [2 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [2 x *] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -432,7 +432,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:51:35: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:51:35: Evaluation criterion node(s):
|
||||
08/16/2016 10:51:35: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 10:51:35: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -484,141 +484,141 @@ BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sampl
|
|||
BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sample 0), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:51:35: Starting minibatch loop.
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.69846765 * 250; EvalErrorPrediction = 0.50400000 * 250; time = 0.0606s; samplesPerSecond = 4125.1
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76129944 * 250; EvalErrorPrediction = 0.46400000 * 250; time = 0.0177s; samplesPerSecond = 14150.7
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72963208 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0497s; samplesPerSecond = 5028.9
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.74041528 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0333s; samplesPerSecond = 7501.9
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70611035 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0223s; samplesPerSecond = 11225.9
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74740723 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0168s; samplesPerSecond = 14876.5
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75085840 * 250; EvalErrorPrediction = 0.40400000 * 250; time = 0.0169s; samplesPerSecond = 14758.8
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78210742 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0170s; samplesPerSecond = 14729.3
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70286572 * 250; EvalErrorPrediction = 0.47200000 * 250; time = 0.0263s; samplesPerSecond = 9508.6
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69580322 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0247s; samplesPerSecond = 10135.4
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70703613 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0170s; samplesPerSecond = 14700.7
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74512988 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0169s; samplesPerSecond = 14772.8
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70837598 * 250; EvalErrorPrediction = 0.47600000 * 250; time = 0.0168s; samplesPerSecond = 14850.9
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69913086 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0173s; samplesPerSecond = 14456.7
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70321875 * 250; EvalErrorPrediction = 0.53600000 * 250; time = 0.0168s; samplesPerSecond = 14899.6
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69290918 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0198s; samplesPerSecond = 12597.0
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74415527 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0141s; samplesPerSecond = 17694.1
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73745117 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0172s; samplesPerSecond = 14513.8
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71849609 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0200s; samplesPerSecond = 12484.4
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71476953 * 250; EvalErrorPrediction = 0.50400000 * 250; time = 0.0284s; samplesPerSecond = 8813.1
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69918457 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0212s; samplesPerSecond = 11786.9
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69749512 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0243s; samplesPerSecond = 10267.4
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70658887 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0140s; samplesPerSecond = 17871.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69760742 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0170s; samplesPerSecond = 14747.5
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69499219 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0169s; samplesPerSecond = 14768.4
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69291211 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0294s; samplesPerSecond = 8497.9
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70718945 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0336s; samplesPerSecond = 7433.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69039453 * 250; EvalErrorPrediction = 0.46800000 * 250; time = 0.0157s; samplesPerSecond = 15957.1
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70257422 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0222s; samplesPerSecond = 11244.0
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71058984 * 250; EvalErrorPrediction = 0.42800000 * 250; time = 0.0151s; samplesPerSecond = 16568.4
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69296875 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0177s; samplesPerSecond = 14113.1
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69641211 * 250; EvalErrorPrediction = 0.51600000 * 250; time = 0.0251s; samplesPerSecond = 9974.1
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69531055 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0164s; samplesPerSecond = 15214.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69090430 * 250; EvalErrorPrediction = 0.47600000 * 250; time = 0.0172s; samplesPerSecond = 14501.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.68339063 * 250; EvalErrorPrediction = 0.48000000 * 250; time = 0.0170s; samplesPerSecond = 14691.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.67383984 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0170s; samplesPerSecond = 14691.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.65904102 * 250; EvalErrorPrediction = 0.26400000 * 250; time = 0.0239s; samplesPerSecond = 10454.6
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.64259766 * 250; EvalErrorPrediction = 0.36000000 * 250; time = 0.0186s; samplesPerSecond = 13465.5
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.60433398 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0196s; samplesPerSecond = 12787.7
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.56497070 * 250; EvalErrorPrediction = 0.11600000 * 250; time = 0.0142s; samplesPerSecond = 17556.2
|
||||
08/16/2016 10:51:36: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70222344 * 10000; EvalErrorPrediction = 0.46820000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.873074s
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.69846765 * 250; EvalClassificationError = 0.50400000 * 250; time = 0.0606s; samplesPerSecond = 4125.1
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76129944 * 250; EvalClassificationError = 0.46400000 * 250; time = 0.0177s; samplesPerSecond = 14150.7
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72963208 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0497s; samplesPerSecond = 5028.9
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.74041528 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0333s; samplesPerSecond = 7501.9
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70611035 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0223s; samplesPerSecond = 11225.9
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74740723 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0168s; samplesPerSecond = 14876.5
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75085840 * 250; EvalClassificationError = 0.40400000 * 250; time = 0.0169s; samplesPerSecond = 14758.8
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78210742 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0170s; samplesPerSecond = 14729.3
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70286572 * 250; EvalClassificationError = 0.47200000 * 250; time = 0.0263s; samplesPerSecond = 9508.6
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69580322 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0247s; samplesPerSecond = 10135.4
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70703613 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0170s; samplesPerSecond = 14700.7
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74512988 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0169s; samplesPerSecond = 14772.8
|
||||
08/16/2016 10:51:35: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70837598 * 250; EvalClassificationError = 0.47600000 * 250; time = 0.0168s; samplesPerSecond = 14850.9
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69913086 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0173s; samplesPerSecond = 14456.7
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70321875 * 250; EvalClassificationError = 0.53600000 * 250; time = 0.0168s; samplesPerSecond = 14899.6
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69290918 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0198s; samplesPerSecond = 12597.0
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74415527 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0141s; samplesPerSecond = 17694.1
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73745117 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0172s; samplesPerSecond = 14513.8
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71849609 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0200s; samplesPerSecond = 12484.4
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71476953 * 250; EvalClassificationError = 0.50400000 * 250; time = 0.0284s; samplesPerSecond = 8813.1
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69918457 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0212s; samplesPerSecond = 11786.9
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69749512 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0243s; samplesPerSecond = 10267.4
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70658887 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0140s; samplesPerSecond = 17871.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69760742 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0170s; samplesPerSecond = 14747.5
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69499219 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0169s; samplesPerSecond = 14768.4
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69291211 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0294s; samplesPerSecond = 8497.9
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70718945 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0336s; samplesPerSecond = 7433.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69039453 * 250; EvalClassificationError = 0.46800000 * 250; time = 0.0157s; samplesPerSecond = 15957.1
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70257422 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0222s; samplesPerSecond = 11244.0
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71058984 * 250; EvalClassificationError = 0.42800000 * 250; time = 0.0151s; samplesPerSecond = 16568.4
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69296875 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0177s; samplesPerSecond = 14113.1
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69641211 * 250; EvalClassificationError = 0.51600000 * 250; time = 0.0251s; samplesPerSecond = 9974.1
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69531055 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0164s; samplesPerSecond = 15214.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69090430 * 250; EvalClassificationError = 0.47600000 * 250; time = 0.0172s; samplesPerSecond = 14501.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.68339063 * 250; EvalClassificationError = 0.48000000 * 250; time = 0.0170s; samplesPerSecond = 14691.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.67383984 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0170s; samplesPerSecond = 14691.2
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.65904102 * 250; EvalClassificationError = 0.26400000 * 250; time = 0.0239s; samplesPerSecond = 10454.6
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.64259766 * 250; EvalClassificationError = 0.36000000 * 250; time = 0.0186s; samplesPerSecond = 13465.5
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.60433398 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0196s; samplesPerSecond = 12787.7
|
||||
08/16/2016 10:51:36: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.56497070 * 250; EvalClassificationError = 0.11600000 * 250; time = 0.0142s; samplesPerSecond = 17556.2
|
||||
08/16/2016 10:51:36: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70222344 * 10000; EvalClassificationError = 0.46820000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.873074s
|
||||
08/16/2016 10:51:36: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095502.258817/Examples/Other/Simple2d_Simple@release_cpu/Models/simple.dnn.1'
|
||||
|
||||
08/16/2016 10:51:36: Starting Epoch 2: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 1: frames [10000..20000] (first sequence at sample 10000), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:51:36: Starting minibatch loop.
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.50663568 * 250; EvalErrorPrediction = 0.15200000 * 250; time = 0.0194s; samplesPerSecond = 12857.4
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.45398022 * 250; EvalErrorPrediction = 0.12000000 * 250; time = 0.0204s; samplesPerSecond = 12253.7
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.37457013 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0168s; samplesPerSecond = 14862.4
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.34124719 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0167s; samplesPerSecond = 14992.5
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.29298340 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0238s; samplesPerSecond = 10498.0
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.27701599 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0332s; samplesPerSecond = 7519.0
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.25128760 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0490s; samplesPerSecond = 5104.9
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.21941431 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0315s; samplesPerSecond = 7933.5
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.22864038 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0145s; samplesPerSecond = 17220.0
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.20533081 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0193s; samplesPerSecond = 12942.6
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.18820288 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0197s; samplesPerSecond = 12660.2
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.17363208 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0207s; samplesPerSecond = 12054.0
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.18979712 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0400s; samplesPerSecond = 6257.7
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.18266016 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0167s; samplesPerSecond = 15002.4
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.18476245 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0167s; samplesPerSecond = 14997.0
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.17951782 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0381s; samplesPerSecond = 6554.3
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.18190771 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0495s; samplesPerSecond = 5048.7
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21016113 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0245s; samplesPerSecond = 10195.3
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.16539111 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0166s; samplesPerSecond = 15091.2
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17295947 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0166s; samplesPerSecond = 15059.3
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13286475 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0134s; samplesPerSecond = 18714.0
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17238135 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0370s; samplesPerSecond = 6753.5
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12533740 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0166s; samplesPerSecond = 15029.5
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21608838 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0306s; samplesPerSecond = 8160.1
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21742236 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0164s; samplesPerSecond = 15279.3
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.17923486 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0174s; samplesPerSecond = 14330.8
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16031250 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0165s; samplesPerSecond = 15119.4
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13486084 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0199s; samplesPerSecond = 12574.8
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16416699 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0162s; samplesPerSecond = 15386.5
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14665625 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0172s; samplesPerSecond = 14556.9
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.14992627 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0190s; samplesPerSecond = 13191.2
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12446338 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0165s; samplesPerSecond = 15123.1
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16560303 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0169s; samplesPerSecond = 14759.7
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14359863 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0163s; samplesPerSecond = 15295.2
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13723389 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0165s; samplesPerSecond = 15156.1
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14104785 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0206s; samplesPerSecond = 12144.8
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15801807 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0259s; samplesPerSecond = 9664.1
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16213721 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0165s; samplesPerSecond = 15138.7
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13545947 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0167s; samplesPerSecond = 15003.3
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15420410 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0197s; samplesPerSecond = 12690.4
|
||||
08/16/2016 10:51:37: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.20252788 * 10000; EvalErrorPrediction = 0.07960000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=0.894097s
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.50663568 * 250; EvalClassificationError = 0.15200000 * 250; time = 0.0194s; samplesPerSecond = 12857.4
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.45398022 * 250; EvalClassificationError = 0.12000000 * 250; time = 0.0204s; samplesPerSecond = 12253.7
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.37457013 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0168s; samplesPerSecond = 14862.4
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.34124719 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0167s; samplesPerSecond = 14992.5
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.29298340 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0238s; samplesPerSecond = 10498.0
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.27701599 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0332s; samplesPerSecond = 7519.0
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.25128760 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0490s; samplesPerSecond = 5104.9
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.21941431 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0315s; samplesPerSecond = 7933.5
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.22864038 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0145s; samplesPerSecond = 17220.0
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.20533081 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0193s; samplesPerSecond = 12942.6
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.18820288 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0197s; samplesPerSecond = 12660.2
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.17363208 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0207s; samplesPerSecond = 12054.0
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.18979712 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0400s; samplesPerSecond = 6257.7
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.18266016 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0167s; samplesPerSecond = 15002.4
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.18476245 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0167s; samplesPerSecond = 14997.0
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.17951782 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0381s; samplesPerSecond = 6554.3
|
||||
08/16/2016 10:51:36: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.18190771 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0495s; samplesPerSecond = 5048.7
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21016113 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0245s; samplesPerSecond = 10195.3
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.16539111 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0166s; samplesPerSecond = 15091.2
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17295947 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0166s; samplesPerSecond = 15059.3
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13286475 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0134s; samplesPerSecond = 18714.0
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17238135 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0370s; samplesPerSecond = 6753.5
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12533740 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0166s; samplesPerSecond = 15029.5
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21608838 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0306s; samplesPerSecond = 8160.1
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21742236 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0164s; samplesPerSecond = 15279.3
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.17923486 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0174s; samplesPerSecond = 14330.8
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16031250 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0165s; samplesPerSecond = 15119.4
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13486084 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0199s; samplesPerSecond = 12574.8
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16416699 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0162s; samplesPerSecond = 15386.5
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14665625 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0172s; samplesPerSecond = 14556.9
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.14992627 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0190s; samplesPerSecond = 13191.2
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12446338 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0165s; samplesPerSecond = 15123.1
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16560303 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0169s; samplesPerSecond = 14759.7
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14359863 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0163s; samplesPerSecond = 15295.2
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13723389 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0165s; samplesPerSecond = 15156.1
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14104785 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0206s; samplesPerSecond = 12144.8
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15801807 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0259s; samplesPerSecond = 9664.1
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16213721 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0165s; samplesPerSecond = 15138.7
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13545947 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0167s; samplesPerSecond = 15003.3
|
||||
08/16/2016 10:51:37: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15420410 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0197s; samplesPerSecond = 12690.4
|
||||
08/16/2016 10:51:37: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.20252788 * 10000; EvalClassificationError = 0.07960000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=0.894097s
|
||||
08/16/2016 10:51:37: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095502.258817/Examples/Other/Simple2d_Simple@release_cpu/Models/simple.dnn.2'
|
||||
|
||||
08/16/2016 10:51:37: Starting Epoch 3: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 2: frames [20000..30000] (first sequence at sample 20000), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:51:37: Starting minibatch loop.
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18365215 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0160s; samplesPerSecond = 15637.7
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12863173 * 250; EvalErrorPrediction = 0.04800000 * 250; time = 0.0195s; samplesPerSecond = 12842.9
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17736676 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0134s; samplesPerSecond = 18714.0
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14110736 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0137s; samplesPerSecond = 18288.2
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16524695 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0221s; samplesPerSecond = 11297.4
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19137244 * 250; EvalErrorPrediction = 0.10800000 * 250; time = 0.0239s; samplesPerSecond = 10451.5
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12233600 * 250; EvalErrorPrediction = 0.04000000 * 250; time = 0.0179s; samplesPerSecond = 13986.0
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16686743 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0229s; samplesPerSecond = 10916.1
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12411963 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0179s; samplesPerSecond = 13940.8
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19959802 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0211s; samplesPerSecond = 11875.4
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14190784 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0180s; samplesPerSecond = 13927.6
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12357324 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0188s; samplesPerSecond = 13270.3
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16388794 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0180s; samplesPerSecond = 13866.5
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19857666 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0179s; samplesPerSecond = 13944.7
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.17161865 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0541s; samplesPerSecond = 4625.3
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13291455 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0204s; samplesPerSecond = 12276.6
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14355762 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0132s; samplesPerSecond = 18926.5
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20757080 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0164s; samplesPerSecond = 15286.8
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19119531 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0288s; samplesPerSecond = 8688.4
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.14750488 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0101s; samplesPerSecond = 24781.9
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15454297 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0204s; samplesPerSecond = 12226.7
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13628662 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0134s; samplesPerSecond = 18693.0
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17363599 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0236s; samplesPerSecond = 10598.6
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14413379 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0100s; samplesPerSecond = 24942.6
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13718579 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0195s; samplesPerSecond = 12810.7
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14220020 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0134s; samplesPerSecond = 18648.4
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16849121 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0164s; samplesPerSecond = 15271.8
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18580225 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0166s; samplesPerSecond = 15018.6
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16339307 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0223s; samplesPerSecond = 11232.4
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15557813 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0133s; samplesPerSecond = 18785.7
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18845215 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0163s; samplesPerSecond = 15311.1
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13286035 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0170s; samplesPerSecond = 14677.4
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14664014 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0164s; samplesPerSecond = 15248.6
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13965381 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0143s; samplesPerSecond = 17474.0
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20020557 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0196s; samplesPerSecond = 12779.2
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12576953 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0197s; samplesPerSecond = 12707.1
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18509766 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0422s; samplesPerSecond = 5925.9
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15134277 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0391s; samplesPerSecond = 6392.4
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11977588 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0355s; samplesPerSecond = 7032.9
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.13046729 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0726s; samplesPerSecond = 3443.6
|
||||
08/16/2016 10:51:38: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15760303 * 10000; EvalErrorPrediction = 0.07280000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=0.876577s
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18365215 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0160s; samplesPerSecond = 15637.7
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12863173 * 250; EvalClassificationError = 0.04800000 * 250; time = 0.0195s; samplesPerSecond = 12842.9
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17736676 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0134s; samplesPerSecond = 18714.0
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14110736 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0137s; samplesPerSecond = 18288.2
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16524695 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0221s; samplesPerSecond = 11297.4
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19137244 * 250; EvalClassificationError = 0.10800000 * 250; time = 0.0239s; samplesPerSecond = 10451.5
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12233600 * 250; EvalClassificationError = 0.04000000 * 250; time = 0.0179s; samplesPerSecond = 13986.0
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16686743 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0229s; samplesPerSecond = 10916.1
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12411963 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0179s; samplesPerSecond = 13940.8
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19959802 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0211s; samplesPerSecond = 11875.4
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14190784 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0180s; samplesPerSecond = 13927.6
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12357324 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0188s; samplesPerSecond = 13270.3
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16388794 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0180s; samplesPerSecond = 13866.5
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19857666 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0179s; samplesPerSecond = 13944.7
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.17161865 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0541s; samplesPerSecond = 4625.3
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13291455 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0204s; samplesPerSecond = 12276.6
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14355762 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0132s; samplesPerSecond = 18926.5
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20757080 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0164s; samplesPerSecond = 15286.8
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19119531 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0288s; samplesPerSecond = 8688.4
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.14750488 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0101s; samplesPerSecond = 24781.9
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15454297 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0204s; samplesPerSecond = 12226.7
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13628662 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0134s; samplesPerSecond = 18693.0
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17363599 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0236s; samplesPerSecond = 10598.6
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14413379 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0100s; samplesPerSecond = 24942.6
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13718579 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0195s; samplesPerSecond = 12810.7
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14220020 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0134s; samplesPerSecond = 18648.4
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16849121 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0164s; samplesPerSecond = 15271.8
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18580225 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0166s; samplesPerSecond = 15018.6
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16339307 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0223s; samplesPerSecond = 11232.4
|
||||
08/16/2016 10:51:37: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15557813 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0133s; samplesPerSecond = 18785.7
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18845215 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0163s; samplesPerSecond = 15311.1
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13286035 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0170s; samplesPerSecond = 14677.4
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14664014 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0164s; samplesPerSecond = 15248.6
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13965381 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0143s; samplesPerSecond = 17474.0
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20020557 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0196s; samplesPerSecond = 12779.2
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12576953 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0197s; samplesPerSecond = 12707.1
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18509766 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0422s; samplesPerSecond = 5925.9
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15134277 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0391s; samplesPerSecond = 6392.4
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11977588 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0355s; samplesPerSecond = 7032.9
|
||||
08/16/2016 10:51:38: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.13046729 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0726s; samplesPerSecond = 3443.6
|
||||
08/16/2016 10:51:38: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15760303 * 10000; EvalClassificationError = 0.07280000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=0.876577s
|
||||
08/16/2016 10:51:38: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095502.258817/Examples/Other/Simple2d_Simple@release_cpu/Models/simple.dnn'
|
||||
08/16/2016 10:51:38: CNTKCommandTrainEnd: Simple_Demo_Train
|
||||
|
||||
|
|
@ -636,7 +636,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -665,7 +665,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *1] -> [2 x 1 x *1]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *1], [2 x 1] -> [2 x 1 x *1]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *1] -> [2 x 1 x *1]
|
||||
Validating --> Prior = Mean (labels) : [2 x *1] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -693,8 +693,8 @@ Memory Sharing: Out of 25 matrices, 2 are shared as 1, and 23 are not shared.
|
|||
ScaledLogLikelihood : [2 x 1 x *1] }
|
||||
|
||||
BlockRandomizer::StartEpoch: epoch 0: frames [0..603] (first sequence at sample 0), data subset 0 of 1
|
||||
08/16/2016 10:51:38: Minibatch[1-1]: EvalErrorPrediction = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10807832 * 603
|
||||
08/16/2016 10:51:38: Final Results: Minibatch[1-1]: EvalErrorPrediction = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10807832 * 603; perplexity = 1.11413500
|
||||
08/16/2016 10:51:38: Minibatch[1-1]: EvalClassificationError = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10807832 * 603
|
||||
08/16/2016 10:51:38: Final Results: Minibatch[1-1]: EvalClassificationError = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10807832 * 603; perplexity = 1.11413500
|
||||
|
||||
08/16/2016 10:51:38: Action "test" complete.
|
||||
|
||||
|
|
@ -710,7 +710,7 @@ Post-processing network...
|
|||
|
||||
8 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -740,7 +740,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *2] -> [2 x 1 x *2]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *2], [2 x 1] -> [2 x 1 x *2]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *2] -> [2 x 1 x *2]
|
||||
Validating --> Prior = Mean (labels) : [2 x *2] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -764,7 +764,7 @@ Allocating matrices for forward and/or backward propagation.
|
|||
Memory Sharing: Out of 25 matrices, 3 are shared as 1, and 22 are not shared.
|
||||
|
||||
{ CrossEntropyWithSoftmax : [1]
|
||||
EvalErrorPrediction : [1]
|
||||
EvalClassificationError : [1]
|
||||
ScaledLogLikelihood : [2 x 1 x *2] }
|
||||
|
||||
Minibatch[0]: ActualMBSize = 603
|
||||
|
|
|
|||
|
|
@ -71,7 +71,7 @@ Simple_Demo_Train = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -170,7 +170,7 @@ Simple_Demo_Train = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -313,7 +313,7 @@ configparameters: Simple.cntk:Simple_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -381,7 +381,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -410,7 +410,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *] -> [2 x 1 x *]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *], [2 x 1] -> [2 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *] -> [2 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [2 x *] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -433,7 +433,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:51:40: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:51:40: Evaluation criterion node(s):
|
||||
08/16/2016 10:51:40: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 10:51:40: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -485,141 +485,141 @@ BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sampl
|
|||
BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sample 0), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:51:40: Starting minibatch loop.
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70124231 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0078s; samplesPerSecond = 32034.9
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76372424 * 250; EvalErrorPrediction = 0.46400000 * 250; time = 0.0064s; samplesPerSecond = 38892.3
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72703027 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0064s; samplesPerSecond = 39166.5
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73895923 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0064s; samplesPerSecond = 38886.3
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70621924 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0065s; samplesPerSecond = 38759.7
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74767041 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0065s; samplesPerSecond = 38753.7
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75094434 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0064s; samplesPerSecond = 38989.4
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78058936 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0064s; samplesPerSecond = 38922.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70407129 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0064s; samplesPerSecond = 39265.0
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69555762 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0064s; samplesPerSecond = 38922.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70626123 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0064s; samplesPerSecond = 38844.0
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74540430 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0064s; samplesPerSecond = 39178.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70824414 * 250; EvalErrorPrediction = 0.47600000 * 250; time = 0.0064s; samplesPerSecond = 39209.5
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69895020 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0064s; samplesPerSecond = 38886.3
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70353223 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0065s; samplesPerSecond = 38669.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69346387 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0064s; samplesPerSecond = 38989.4
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74449902 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0064s; samplesPerSecond = 38886.3
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73767969 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0064s; samplesPerSecond = 39025.9
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71876855 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0063s; samplesPerSecond = 39594.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71509473 * 250; EvalErrorPrediction = 0.50400000 * 250; time = 0.0064s; samplesPerSecond = 39271.1
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69956152 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0064s; samplesPerSecond = 38886.3
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0062s; samplesPerSecond = 40303.1
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70736035 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0063s; samplesPerSecond = 39563.2
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69820508 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0062s; samplesPerSecond = 40512.1
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69537109 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0063s; samplesPerSecond = 39432.2
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69347266 * 250; EvalErrorPrediction = 0.51200000 * 250; time = 0.0062s; samplesPerSecond = 40492.4
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70801172 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0061s; samplesPerSecond = 40909.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69131641 * 250; EvalErrorPrediction = 0.46800000 * 250; time = 0.0062s; samplesPerSecond = 40257.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70370312 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0062s; samplesPerSecond = 40270.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71200195 * 250; EvalErrorPrediction = 0.43600000 * 250; time = 0.0061s; samplesPerSecond = 40909.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69506836 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0062s; samplesPerSecond = 40577.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69935352 * 250; EvalErrorPrediction = 0.51600000 * 250; time = 0.0061s; samplesPerSecond = 40889.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69887109 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0062s; samplesPerSecond = 40440.0
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69604492 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0062s; samplesPerSecond = 40512.1
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.69011719 * 250; EvalErrorPrediction = 0.48800000 * 250; time = 0.0062s; samplesPerSecond = 40617.4
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.68419531 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0061s; samplesPerSecond = 40783.0
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.67551367 * 250; EvalErrorPrediction = 0.32400000 * 250; time = 0.0063s; samplesPerSecond = 39904.2
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.67028516 * 250; EvalErrorPrediction = 0.40000000 * 250; time = 0.0062s; samplesPerSecond = 40044.9
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.65152734 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0062s; samplesPerSecond = 40630.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.63594727 * 250; EvalErrorPrediction = 0.22000000 * 250; time = 0.0062s; samplesPerSecond = 40283.6
|
||||
08/16/2016 10:51:40: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70729233 * 10000; EvalErrorPrediction = 0.47740000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.256818s
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70124231 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0078s; samplesPerSecond = 32034.9
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76372424 * 250; EvalClassificationError = 0.46400000 * 250; time = 0.0064s; samplesPerSecond = 38892.3
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72703027 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0064s; samplesPerSecond = 39166.5
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73895923 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0064s; samplesPerSecond = 38886.3
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70621924 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0065s; samplesPerSecond = 38759.7
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74767041 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0065s; samplesPerSecond = 38753.7
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75094434 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0064s; samplesPerSecond = 38989.4
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78058936 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0064s; samplesPerSecond = 38922.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70407129 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0064s; samplesPerSecond = 39265.0
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69555762 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0064s; samplesPerSecond = 38922.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70626123 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0064s; samplesPerSecond = 38844.0
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74540430 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0064s; samplesPerSecond = 39178.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70824414 * 250; EvalClassificationError = 0.47600000 * 250; time = 0.0064s; samplesPerSecond = 39209.5
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69895020 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0064s; samplesPerSecond = 38886.3
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70353223 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0065s; samplesPerSecond = 38669.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69346387 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0064s; samplesPerSecond = 38989.4
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74449902 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0064s; samplesPerSecond = 38886.3
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73767969 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0064s; samplesPerSecond = 39025.9
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71876855 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0063s; samplesPerSecond = 39594.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71509473 * 250; EvalClassificationError = 0.50400000 * 250; time = 0.0064s; samplesPerSecond = 39271.1
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69956152 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0064s; samplesPerSecond = 38886.3
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0062s; samplesPerSecond = 40303.1
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70736035 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0063s; samplesPerSecond = 39563.2
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69820508 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0062s; samplesPerSecond = 40512.1
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69537109 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0063s; samplesPerSecond = 39432.2
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69347266 * 250; EvalClassificationError = 0.51200000 * 250; time = 0.0062s; samplesPerSecond = 40492.4
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70801172 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0061s; samplesPerSecond = 40909.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69131641 * 250; EvalClassificationError = 0.46800000 * 250; time = 0.0062s; samplesPerSecond = 40257.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70370312 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0062s; samplesPerSecond = 40270.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71200195 * 250; EvalClassificationError = 0.43600000 * 250; time = 0.0061s; samplesPerSecond = 40909.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69506836 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0062s; samplesPerSecond = 40577.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69935352 * 250; EvalClassificationError = 0.51600000 * 250; time = 0.0061s; samplesPerSecond = 40889.8
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69887109 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0062s; samplesPerSecond = 40440.0
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69604492 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0062s; samplesPerSecond = 40512.1
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.69011719 * 250; EvalClassificationError = 0.48800000 * 250; time = 0.0062s; samplesPerSecond = 40617.4
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.68419531 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0061s; samplesPerSecond = 40783.0
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.67551367 * 250; EvalClassificationError = 0.32400000 * 250; time = 0.0063s; samplesPerSecond = 39904.2
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.67028516 * 250; EvalClassificationError = 0.40000000 * 250; time = 0.0062s; samplesPerSecond = 40044.9
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.65152734 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0062s; samplesPerSecond = 40630.6
|
||||
08/16/2016 10:51:40: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.63594727 * 250; EvalClassificationError = 0.22000000 * 250; time = 0.0062s; samplesPerSecond = 40283.6
|
||||
08/16/2016 10:51:40: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70729233 * 10000; EvalClassificationError = 0.47740000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.256818s
|
||||
08/16/2016 10:51:40: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095502.258817/Examples/Other/Simple2d_Simple@release_gpu/Models/simple.dnn.1'
|
||||
|
||||
08/16/2016 10:51:40: Starting Epoch 2: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 1: frames [10000..20000] (first sequence at sample 10000), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:51:40: Starting minibatch loop.
|
||||
08/16/2016 10:51:40: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.61492108 * 250; EvalErrorPrediction = 0.26800000 * 250; time = 0.0064s; samplesPerSecond = 38801.8
|
||||
08/16/2016 10:51:40: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.59171271 * 250; EvalErrorPrediction = 0.28400000 * 250; time = 0.0063s; samplesPerSecond = 39923.3
|
||||
08/16/2016 10:51:40: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.53591638 * 250; EvalErrorPrediction = 0.20000000 * 250; time = 0.0062s; samplesPerSecond = 40122.0
|
||||
08/16/2016 10:51:40: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.51872742 * 250; EvalErrorPrediction = 0.14000000 * 250; time = 0.0062s; samplesPerSecond = 40479.3
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.48384375 * 250; EvalErrorPrediction = 0.12400000 * 250; time = 0.0062s; samplesPerSecond = 40109.1
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.43163501 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0062s; samplesPerSecond = 40128.4
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.38970386 * 250; EvalErrorPrediction = 0.12400000 * 250; time = 0.0063s; samplesPerSecond = 39733.0
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.33681616 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0062s; samplesPerSecond = 40044.9
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.31352393 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0062s; samplesPerSecond = 40525.2
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.26829492 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0062s; samplesPerSecond = 40270.6
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.24240820 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0063s; samplesPerSecond = 39531.9
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.21015820 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0062s; samplesPerSecond = 40012.8
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.22358789 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0061s; samplesPerSecond = 40856.3
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20496631 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0061s; samplesPerSecond = 40756.4
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.20070508 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0062s; samplesPerSecond = 40643.8
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.19224707 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0061s; samplesPerSecond = 40896.5
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.19326562 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0061s; samplesPerSecond = 40789.7
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21712451 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0061s; samplesPerSecond = 40883.1
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.17562354 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0061s; samplesPerSecond = 40869.7
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.18186035 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0062s; samplesPerSecond = 40577.8
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.14065234 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0062s; samplesPerSecond = 40212.3
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17710254 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0065s; samplesPerSecond = 38711.7
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.13001953 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0064s; samplesPerSecond = 38819.9
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21622949 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0063s; samplesPerSecond = 39613.4
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21902246 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0063s; samplesPerSecond = 39904.2
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18068799 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0064s; samplesPerSecond = 39332.9
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16232471 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 39160.4
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13792139 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0063s; samplesPerSecond = 39607.1
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16526709 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 39080.8
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14743457 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0063s; samplesPerSecond = 39619.7
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15089160 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0064s; samplesPerSecond = 39339.1
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12636230 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0063s; samplesPerSecond = 39834.3
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16735547 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0063s; samplesPerSecond = 39382.5
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14530957 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0064s; samplesPerSecond = 39044.2
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13859570 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0063s; samplesPerSecond = 39638.5
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14215234 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0064s; samplesPerSecond = 39351.5
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15903027 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0064s; samplesPerSecond = 39203.4
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16232520 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 39191.1
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13596484 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0064s; samplesPerSecond = 39099.2
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15469434 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 38965.1
|
||||
08/16/2016 10:51:41: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.24215964 * 10000; EvalErrorPrediction = 0.09440000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=0.253663s
|
||||
08/16/2016 10:51:40: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.61492108 * 250; EvalClassificationError = 0.26800000 * 250; time = 0.0064s; samplesPerSecond = 38801.8
|
||||
08/16/2016 10:51:40: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.59171271 * 250; EvalClassificationError = 0.28400000 * 250; time = 0.0063s; samplesPerSecond = 39923.3
|
||||
08/16/2016 10:51:40: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.53591638 * 250; EvalClassificationError = 0.20000000 * 250; time = 0.0062s; samplesPerSecond = 40122.0
|
||||
08/16/2016 10:51:40: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.51872742 * 250; EvalClassificationError = 0.14000000 * 250; time = 0.0062s; samplesPerSecond = 40479.3
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.48384375 * 250; EvalClassificationError = 0.12400000 * 250; time = 0.0062s; samplesPerSecond = 40109.1
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.43163501 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0062s; samplesPerSecond = 40128.4
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.38970386 * 250; EvalClassificationError = 0.12400000 * 250; time = 0.0063s; samplesPerSecond = 39733.0
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.33681616 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0062s; samplesPerSecond = 40044.9
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.31352393 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0062s; samplesPerSecond = 40525.2
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.26829492 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0062s; samplesPerSecond = 40270.6
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.24240820 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0063s; samplesPerSecond = 39531.9
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.21015820 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0062s; samplesPerSecond = 40012.8
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.22358789 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0061s; samplesPerSecond = 40856.3
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20496631 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0061s; samplesPerSecond = 40756.4
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.20070508 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0062s; samplesPerSecond = 40643.8
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.19224707 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0061s; samplesPerSecond = 40896.5
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.19326562 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0061s; samplesPerSecond = 40789.7
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21712451 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0061s; samplesPerSecond = 40883.1
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.17562354 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0061s; samplesPerSecond = 40869.7
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.18186035 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0062s; samplesPerSecond = 40577.8
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.14065234 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0062s; samplesPerSecond = 40212.3
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17710254 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0065s; samplesPerSecond = 38711.7
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.13001953 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0064s; samplesPerSecond = 38819.9
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21622949 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0063s; samplesPerSecond = 39613.4
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21902246 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0063s; samplesPerSecond = 39904.2
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18068799 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0064s; samplesPerSecond = 39332.9
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16232471 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 39160.4
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13792139 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0063s; samplesPerSecond = 39607.1
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16526709 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 39080.8
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14743457 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0063s; samplesPerSecond = 39619.7
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15089160 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0064s; samplesPerSecond = 39339.1
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12636230 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0063s; samplesPerSecond = 39834.3
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16735547 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0063s; samplesPerSecond = 39382.5
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14530957 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0064s; samplesPerSecond = 39044.2
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13859570 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0063s; samplesPerSecond = 39638.5
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14215234 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0064s; samplesPerSecond = 39351.5
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15903027 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0064s; samplesPerSecond = 39203.4
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16232520 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 39191.1
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13596484 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0064s; samplesPerSecond = 39099.2
|
||||
08/16/2016 10:51:41: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15469434 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 38965.1
|
||||
08/16/2016 10:51:41: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.24215964 * 10000; EvalClassificationError = 0.09440000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=0.253663s
|
||||
08/16/2016 10:51:41: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095502.258817/Examples/Other/Simple2d_Simple@release_gpu/Models/simple.dnn.2'
|
||||
|
||||
08/16/2016 10:51:41: Starting Epoch 3: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 2: frames [20000..30000] (first sequence at sample 20000), data subset 0 of 1
|
||||
|
||||
08/16/2016 10:51:41: Starting minibatch loop.
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18305315 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0064s; samplesPerSecond = 38880.2
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12945729 * 250; EvalErrorPrediction = 0.04800000 * 250; time = 0.0063s; samplesPerSecond = 39980.8
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17735931 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0065s; samplesPerSecond = 38729.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14128339 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0064s; samplesPerSecond = 39013.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16558209 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0064s; samplesPerSecond = 39080.8
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19102692 * 250; EvalErrorPrediction = 0.10800000 * 250; time = 0.0065s; samplesPerSecond = 38627.9
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12279083 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0064s; samplesPerSecond = 39001.6
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16642798 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0065s; samplesPerSecond = 38314.2
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12386572 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0064s; samplesPerSecond = 38844.0
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19928418 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0065s; samplesPerSecond = 38681.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14213635 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0064s; samplesPerSecond = 38898.4
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12377087 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0062s; samplesPerSecond = 40032.0
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16361621 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0063s; samplesPerSecond = 39789.9
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19886914 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0063s; samplesPerSecond = 39821.6
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.17207544 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0063s; samplesPerSecond = 39968.0
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13323437 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0063s; samplesPerSecond = 39663.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14397510 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0063s; samplesPerSecond = 39866.1
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20777515 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0063s; samplesPerSecond = 39980.8
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19094092 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0062s; samplesPerSecond = 40057.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.14731372 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0062s; samplesPerSecond = 40038.4
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15483569 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 39252.6
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13625415 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0065s; samplesPerSecond = 38491.1
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17354004 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0063s; samplesPerSecond = 39942.5
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14408350 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 39013.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13720044 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0062s; samplesPerSecond = 40025.6
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14236426 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0062s; samplesPerSecond = 40019.2
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16857861 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0063s; samplesPerSecond = 39847.0
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18606982 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0062s; samplesPerSecond = 40381.2
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16334619 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0062s; samplesPerSecond = 40199.4
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15598535 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0063s; samplesPerSecond = 39827.9
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18848584 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0062s; samplesPerSecond = 40238.2
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13281348 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0063s; samplesPerSecond = 39669.9
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14679150 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0063s; samplesPerSecond = 39419.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13977344 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0063s; samplesPerSecond = 39726.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20015137 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0062s; samplesPerSecond = 40244.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12582129 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0063s; samplesPerSecond = 39388.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18500098 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0062s; samplesPerSecond = 40051.3
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15147754 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0062s; samplesPerSecond = 40057.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11988379 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0063s; samplesPerSecond = 39827.9
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.13059082 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0064s; samplesPerSecond = 39345.3
|
||||
08/16/2016 10:51:41: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15767216 * 10000; EvalErrorPrediction = 0.07330000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=0.255461s
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18305315 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0064s; samplesPerSecond = 38880.2
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12945729 * 250; EvalClassificationError = 0.04800000 * 250; time = 0.0063s; samplesPerSecond = 39980.8
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17735931 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0065s; samplesPerSecond = 38729.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14128339 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0064s; samplesPerSecond = 39013.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16558209 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0064s; samplesPerSecond = 39080.8
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19102692 * 250; EvalClassificationError = 0.10800000 * 250; time = 0.0065s; samplesPerSecond = 38627.9
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12279083 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0064s; samplesPerSecond = 39001.6
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16642798 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0065s; samplesPerSecond = 38314.2
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12386572 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0064s; samplesPerSecond = 38844.0
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19928418 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0065s; samplesPerSecond = 38681.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14213635 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0064s; samplesPerSecond = 38898.4
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12377087 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0062s; samplesPerSecond = 40032.0
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16361621 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0063s; samplesPerSecond = 39789.9
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19886914 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0063s; samplesPerSecond = 39821.6
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.17207544 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0063s; samplesPerSecond = 39968.0
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13323437 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0063s; samplesPerSecond = 39663.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14397510 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0063s; samplesPerSecond = 39866.1
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20777515 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0063s; samplesPerSecond = 39980.8
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19094092 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0062s; samplesPerSecond = 40057.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.14731372 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0062s; samplesPerSecond = 40038.4
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15483569 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 39252.6
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13625415 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0065s; samplesPerSecond = 38491.1
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17354004 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0063s; samplesPerSecond = 39942.5
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14408350 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0064s; samplesPerSecond = 39013.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13720044 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0062s; samplesPerSecond = 40025.6
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14236426 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0062s; samplesPerSecond = 40019.2
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16857861 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0063s; samplesPerSecond = 39847.0
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18606982 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0062s; samplesPerSecond = 40381.2
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16334619 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0062s; samplesPerSecond = 40199.4
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15598535 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0063s; samplesPerSecond = 39827.9
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18848584 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0062s; samplesPerSecond = 40238.2
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13281348 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0063s; samplesPerSecond = 39669.9
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14679150 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0063s; samplesPerSecond = 39419.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13977344 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0063s; samplesPerSecond = 39726.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20015137 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0062s; samplesPerSecond = 40244.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12582129 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0063s; samplesPerSecond = 39388.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18500098 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0062s; samplesPerSecond = 40051.3
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15147754 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0062s; samplesPerSecond = 40057.7
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11988379 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0063s; samplesPerSecond = 39827.9
|
||||
08/16/2016 10:51:41: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.13059082 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0064s; samplesPerSecond = 39345.3
|
||||
08/16/2016 10:51:41: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15767216 * 10000; EvalClassificationError = 0.07330000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=0.255461s
|
||||
08/16/2016 10:51:41: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095502.258817/Examples/Other/Simple2d_Simple@release_gpu/Models/simple.dnn'
|
||||
08/16/2016 10:51:41: CNTKCommandTrainEnd: Simple_Demo_Train
|
||||
|
||||
|
|
@ -637,7 +637,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -666,7 +666,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *1] -> [2 x 1 x *1]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *1], [2 x 1] -> [2 x 1 x *1]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *1] -> [2 x 1 x *1]
|
||||
Validating --> Prior = Mean (labels) : [2 x *1] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -694,8 +694,8 @@ Memory Sharing: Out of 25 matrices, 2 are shared as 1, and 23 are not shared.
|
|||
ScaledLogLikelihood : [2 x 1 x *1] }
|
||||
|
||||
BlockRandomizer::StartEpoch: epoch 0: frames [0..603] (first sequence at sample 0), data subset 0 of 1
|
||||
08/16/2016 10:51:41: Minibatch[1-1]: EvalErrorPrediction = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10845041 * 603
|
||||
08/16/2016 10:51:41: Final Results: Minibatch[1-1]: EvalErrorPrediction = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10845041 * 603; perplexity = 1.11454964
|
||||
08/16/2016 10:51:41: Minibatch[1-1]: EvalClassificationError = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10845041 * 603
|
||||
08/16/2016 10:51:41: Final Results: Minibatch[1-1]: EvalClassificationError = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10845041 * 603; perplexity = 1.11454964
|
||||
|
||||
08/16/2016 10:51:41: Action "test" complete.
|
||||
|
||||
|
|
@ -711,7 +711,7 @@ Post-processing network...
|
|||
|
||||
8 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -741,7 +741,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *2] -> [2 x 1 x *2]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *2], [2 x 1] -> [2 x 1 x *2]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *2] -> [2 x 1 x *2]
|
||||
Validating --> Prior = Mean (labels) : [2 x *2] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -765,7 +765,7 @@ Allocating matrices for forward and/or backward propagation.
|
|||
Memory Sharing: Out of 25 matrices, 3 are shared as 1, and 22 are not shared.
|
||||
|
||||
{ CrossEntropyWithSoftmax : [1]
|
||||
EvalErrorPrediction : [1]
|
||||
EvalClassificationError : [1]
|
||||
ScaledLogLikelihood : [2 x 1 x *2] }
|
||||
|
||||
Minibatch[0]: ActualMBSize = 603
|
||||
|
|
|
|||
|
|
@ -71,7 +71,7 @@ Simple_Demo_Train = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -170,7 +170,7 @@ Simple_Demo_Train = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -313,7 +313,7 @@ configparameters: Simple.cntk:Simple_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -380,7 +380,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -409,7 +409,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *] -> [2 x 1 x *]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *], [2 x 1] -> [2 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *] -> [2 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [2 x *] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -432,7 +432,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:04:16: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:04:16: Evaluation criterion node(s):
|
||||
08/16/2016 03:04:16: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 03:04:16: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -484,141 +484,141 @@ BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sampl
|
|||
BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sample 0), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:04:16: Starting minibatch loop.
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70264496 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0355s; samplesPerSecond = 7041.1
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76483063 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0281s; samplesPerSecond = 8903.5
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72648584 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0269s; samplesPerSecond = 9307.5
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73860254 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0254s; samplesPerSecond = 9858.4
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70622803 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0248s; samplesPerSecond = 10062.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74772852 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0246s; samplesPerSecond = 10142.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75092773 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0253s; samplesPerSecond = 9869.3
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78004932 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0250s; samplesPerSecond = 9983.2
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70444336 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0257s; samplesPerSecond = 9745.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69544189 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0253s; samplesPerSecond = 9889.6
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70595947 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0255s; samplesPerSecond = 9823.2
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74544189 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0250s; samplesPerSecond = 9994.4
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70809961 * 250; EvalErrorPrediction = 0.47600000 * 250; time = 0.0253s; samplesPerSecond = 9888.5
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69884375 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0252s; samplesPerSecond = 9917.5
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70363086 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0257s; samplesPerSecond = 9717.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69351758 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0250s; samplesPerSecond = 9998.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74453613 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0252s; samplesPerSecond = 9901.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73761426 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0247s; samplesPerSecond = 10133.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71868652 * 250; EvalErrorPrediction = 0.48800000 * 250; time = 0.0256s; samplesPerSecond = 9782.1
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71496484 * 250; EvalErrorPrediction = 0.50400000 * 250; time = 0.0249s; samplesPerSecond = 10052.7
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69961230 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0249s; samplesPerSecond = 10036.1
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69760645 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0260s; samplesPerSecond = 9618.3
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70748047 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0256s; samplesPerSecond = 9771.7
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0253s; samplesPerSecond = 9882.6
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69483203 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0256s; samplesPerSecond = 9754.6
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69258203 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0238s; samplesPerSecond = 10503.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70665625 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0245s; samplesPerSecond = 10191.2
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69031445 * 250; EvalErrorPrediction = 0.46800000 * 250; time = 0.0241s; samplesPerSecond = 10352.4
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70169531 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0242s; samplesPerSecond = 10326.3
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71008398 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0238s; samplesPerSecond = 10486.6
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69152930 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0242s; samplesPerSecond = 10347.3
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69522656 * 250; EvalErrorPrediction = 0.51600000 * 250; time = 0.0239s; samplesPerSecond = 10472.1
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69347070 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0243s; samplesPerSecond = 10308.9
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.68888281 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0242s; samplesPerSecond = 10329.7
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.68067578 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0243s; samplesPerSecond = 10280.9
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.66932227 * 250; EvalErrorPrediction = 0.44400000 * 250; time = 0.0242s; samplesPerSecond = 10317.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.65398437 * 250; EvalErrorPrediction = 0.24800000 * 250; time = 0.0237s; samplesPerSecond = 10545.4
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.63662500 * 250; EvalErrorPrediction = 0.32400000 * 250; time = 0.0240s; samplesPerSecond = 10400.6
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.59652344 * 250; EvalErrorPrediction = 0.04800000 * 250; time = 0.0242s; samplesPerSecond = 10346.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.55820898 * 250; EvalErrorPrediction = 0.12000000 * 250; time = 0.0238s; samplesPerSecond = 10488.3
|
||||
08/16/2016 03:04:17: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70129624 * 10000; EvalErrorPrediction = 0.46850000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=1.01068s
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70264496 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0355s; samplesPerSecond = 7041.1
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76483063 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0281s; samplesPerSecond = 8903.5
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72648584 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0269s; samplesPerSecond = 9307.5
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73860254 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0254s; samplesPerSecond = 9858.4
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70622803 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0248s; samplesPerSecond = 10062.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74772852 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0246s; samplesPerSecond = 10142.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75092773 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0253s; samplesPerSecond = 9869.3
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78004932 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0250s; samplesPerSecond = 9983.2
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70444336 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0257s; samplesPerSecond = 9745.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69544189 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0253s; samplesPerSecond = 9889.6
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70595947 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0255s; samplesPerSecond = 9823.2
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74544189 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0250s; samplesPerSecond = 9994.4
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70809961 * 250; EvalClassificationError = 0.47600000 * 250; time = 0.0253s; samplesPerSecond = 9888.5
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69884375 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0252s; samplesPerSecond = 9917.5
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70363086 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0257s; samplesPerSecond = 9717.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69351758 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0250s; samplesPerSecond = 9998.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74453613 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0252s; samplesPerSecond = 9901.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73761426 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0247s; samplesPerSecond = 10133.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71868652 * 250; EvalClassificationError = 0.48800000 * 250; time = 0.0256s; samplesPerSecond = 9782.1
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71496484 * 250; EvalClassificationError = 0.50400000 * 250; time = 0.0249s; samplesPerSecond = 10052.7
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69961230 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0249s; samplesPerSecond = 10036.1
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69760645 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0260s; samplesPerSecond = 9618.3
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70748047 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0256s; samplesPerSecond = 9771.7
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0253s; samplesPerSecond = 9882.6
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69483203 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0256s; samplesPerSecond = 9754.6
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69258203 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0238s; samplesPerSecond = 10503.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70665625 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0245s; samplesPerSecond = 10191.2
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69031445 * 250; EvalClassificationError = 0.46800000 * 250; time = 0.0241s; samplesPerSecond = 10352.4
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70169531 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0242s; samplesPerSecond = 10326.3
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71008398 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0238s; samplesPerSecond = 10486.6
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69152930 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0242s; samplesPerSecond = 10347.3
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69522656 * 250; EvalClassificationError = 0.51600000 * 250; time = 0.0239s; samplesPerSecond = 10472.1
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69347070 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0243s; samplesPerSecond = 10308.9
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.68888281 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0242s; samplesPerSecond = 10329.7
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.68067578 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0243s; samplesPerSecond = 10280.9
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.66932227 * 250; EvalClassificationError = 0.44400000 * 250; time = 0.0242s; samplesPerSecond = 10317.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.65398437 * 250; EvalClassificationError = 0.24800000 * 250; time = 0.0237s; samplesPerSecond = 10545.4
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.63662500 * 250; EvalClassificationError = 0.32400000 * 250; time = 0.0240s; samplesPerSecond = 10400.6
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.59652344 * 250; EvalClassificationError = 0.04800000 * 250; time = 0.0242s; samplesPerSecond = 10346.8
|
||||
08/16/2016 03:04:17: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.55820898 * 250; EvalClassificationError = 0.12000000 * 250; time = 0.0238s; samplesPerSecond = 10488.3
|
||||
08/16/2016 03:04:17: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70129624 * 10000; EvalClassificationError = 0.46850000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=1.01068s
|
||||
08/16/2016 03:04:17: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030038.674053\Examples\Other\Simple2d_Simple@release_cpu/Models/simple.dnn.1'
|
||||
|
||||
08/16/2016 03:04:18: Starting Epoch 2: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 1: frames [10000..20000] (first sequence at sample 10000), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:04:18: Starting minibatch loop.
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.50449603 * 250; EvalErrorPrediction = 0.14800000 * 250; time = 0.0230s; samplesPerSecond = 10862.5
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.45593445 * 250; EvalErrorPrediction = 0.12800000 * 250; time = 0.0229s; samplesPerSecond = 10916.6
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.38063666 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0230s; samplesPerSecond = 10880.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.35197192 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0227s; samplesPerSecond = 11005.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.30828760 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0229s; samplesPerSecond = 10918.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.29232886 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0228s; samplesPerSecond = 10979.4
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.26675781 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0230s; samplesPerSecond = 10878.6
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.23178394 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0230s; samplesPerSecond = 10857.3
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.23917383 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0228s; samplesPerSecond = 10954.3
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.21675732 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0227s; samplesPerSecond = 11001.6
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.19885376 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0230s; samplesPerSecond = 10854.5
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.18136646 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0232s; samplesPerSecond = 10786.6
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.19802368 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0231s; samplesPerSecond = 10826.7
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.18948218 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0227s; samplesPerSecond = 10990.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.18990088 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0230s; samplesPerSecond = 10861.1
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.18491504 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0232s; samplesPerSecond = 10772.1
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.18686621 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0232s; samplesPerSecond = 10788.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21271729 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0232s; samplesPerSecond = 10780.5
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.16924951 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0225s; samplesPerSecond = 11127.4
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17609473 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0231s; samplesPerSecond = 10845.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13717920 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0229s; samplesPerSecond = 10921.8
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17546387 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0233s; samplesPerSecond = 10708.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12864746 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0228s; samplesPerSecond = 10944.8
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21596680 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0231s; samplesPerSecond = 10832.8
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21857666 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0228s; samplesPerSecond = 10946.7
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18096436 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0239s; samplesPerSecond = 10463.8
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16132373 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0230s; samplesPerSecond = 10881.4
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13699268 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0228s; samplesPerSecond = 10960.6
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16551953 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0229s; samplesPerSecond = 10909.4
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14865527 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0238s; samplesPerSecond = 10483.1
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15119824 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0226s; samplesPerSecond = 11060.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12673340 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0230s; samplesPerSecond = 10887.1
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16551514 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0231s; samplesPerSecond = 10808.9
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14445264 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0233s; samplesPerSecond = 10734.2
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13810986 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0230s; samplesPerSecond = 10880.4
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14219189 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0225s; samplesPerSecond = 11107.2
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15920459 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0225s; samplesPerSecond = 11113.1
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16245654 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0238s; samplesPerSecond = 10512.2
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13554053 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0228s; samplesPerSecond = 10988.5
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15504346 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0228s; samplesPerSecond = 10968.3
|
||||
08/16/2016 03:04:18: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.20713335 * 10000; EvalErrorPrediction = 0.08030000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=0.921702s
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.50449603 * 250; EvalClassificationError = 0.14800000 * 250; time = 0.0230s; samplesPerSecond = 10862.5
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.45593445 * 250; EvalClassificationError = 0.12800000 * 250; time = 0.0229s; samplesPerSecond = 10916.6
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.38063666 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0230s; samplesPerSecond = 10880.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.35197192 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0227s; samplesPerSecond = 11005.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.30828760 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0229s; samplesPerSecond = 10918.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.29232886 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0228s; samplesPerSecond = 10979.4
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.26675781 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0230s; samplesPerSecond = 10878.6
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.23178394 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0230s; samplesPerSecond = 10857.3
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.23917383 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0228s; samplesPerSecond = 10954.3
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.21675732 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0227s; samplesPerSecond = 11001.6
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.19885376 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0230s; samplesPerSecond = 10854.5
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.18136646 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0232s; samplesPerSecond = 10786.6
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.19802368 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0231s; samplesPerSecond = 10826.7
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.18948218 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0227s; samplesPerSecond = 10990.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.18990088 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0230s; samplesPerSecond = 10861.1
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.18491504 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0232s; samplesPerSecond = 10772.1
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.18686621 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0232s; samplesPerSecond = 10788.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21271729 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0232s; samplesPerSecond = 10780.5
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.16924951 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0225s; samplesPerSecond = 11127.4
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.17609473 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0231s; samplesPerSecond = 10845.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.13717920 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0229s; samplesPerSecond = 10921.8
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17546387 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0233s; samplesPerSecond = 10708.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.12864746 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0228s; samplesPerSecond = 10944.8
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21596680 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0231s; samplesPerSecond = 10832.8
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21857666 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0228s; samplesPerSecond = 10946.7
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18096436 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0239s; samplesPerSecond = 10463.8
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16132373 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0230s; samplesPerSecond = 10881.4
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13699268 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0228s; samplesPerSecond = 10960.6
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16551953 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0229s; samplesPerSecond = 10909.4
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14865527 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0238s; samplesPerSecond = 10483.1
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15119824 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0226s; samplesPerSecond = 11060.0
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12673340 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0230s; samplesPerSecond = 10887.1
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16551514 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0231s; samplesPerSecond = 10808.9
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14445264 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0233s; samplesPerSecond = 10734.2
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13810986 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0230s; samplesPerSecond = 10880.4
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14219189 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0225s; samplesPerSecond = 11107.2
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15920459 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0225s; samplesPerSecond = 11113.1
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16245654 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0238s; samplesPerSecond = 10512.2
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13554053 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0228s; samplesPerSecond = 10988.5
|
||||
08/16/2016 03:04:18: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15504346 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0228s; samplesPerSecond = 10968.3
|
||||
08/16/2016 03:04:18: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.20713335 * 10000; EvalClassificationError = 0.08030000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=0.921702s
|
||||
08/16/2016 03:04:18: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030038.674053\Examples\Other\Simple2d_Simple@release_cpu/Models/simple.dnn.2'
|
||||
|
||||
08/16/2016 03:04:18: Starting Epoch 3: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 2: frames [20000..30000] (first sequence at sample 20000), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:04:18: Starting minibatch loop.
|
||||
08/16/2016 03:04:18: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18297285 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0231s; samplesPerSecond = 10833.8
|
||||
08/16/2016 03:04:18: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12934721 * 250; EvalErrorPrediction = 0.04800000 * 250; time = 0.0230s; samplesPerSecond = 10872.4
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17702411 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0225s; samplesPerSecond = 11110.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14030841 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0228s; samplesPerSecond = 10941.4
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16429517 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0223s; samplesPerSecond = 11187.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19154443 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0229s; samplesPerSecond = 10924.7
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12275391 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0244s; samplesPerSecond = 10245.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16801855 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0228s; samplesPerSecond = 10987.6
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12472571 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0230s; samplesPerSecond = 10877.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19939526 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0229s; samplesPerSecond = 10895.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14222791 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0227s; samplesPerSecond = 10995.3
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12374048 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0228s; samplesPerSecond = 10962.0
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16442969 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0224s; samplesPerSecond = 11142.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19837036 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0227s; samplesPerSecond = 11003.0
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.17180200 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0235s; samplesPerSecond = 10638.3
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13326343 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0229s; samplesPerSecond = 10936.6
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14289917 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0233s; samplesPerSecond = 10727.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20692944 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0231s; samplesPerSecond = 10827.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19077197 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0231s; samplesPerSecond = 10817.4
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.14746069 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0227s; samplesPerSecond = 11027.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15464526 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0227s; samplesPerSecond = 11007.9
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13673071 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0229s; samplesPerSecond = 10923.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17348853 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0226s; samplesPerSecond = 11065.4
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14420581 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0233s; samplesPerSecond = 10725.5
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13774097 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0228s; samplesPerSecond = 10975.0
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14177905 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0225s; samplesPerSecond = 11100.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16864648 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0229s; samplesPerSecond = 10928.0
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18513623 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0228s; samplesPerSecond = 10968.3
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16393555 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0226s; samplesPerSecond = 11067.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15467676 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0228s; samplesPerSecond = 10969.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18951318 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0223s; samplesPerSecond = 11207.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13329639 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0232s; samplesPerSecond = 10793.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14604785 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0229s; samplesPerSecond = 10894.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13938086 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0229s; samplesPerSecond = 10896.6
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.19969873 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0226s; samplesPerSecond = 11040.9
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12584180 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0228s; samplesPerSecond = 10949.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18373438 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0232s; samplesPerSecond = 10780.0
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15064795 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0234s; samplesPerSecond = 10683.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11991260 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0229s; samplesPerSecond = 10911.3
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.13070557 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0230s; samplesPerSecond = 10857.8
|
||||
08/16/2016 03:04:19: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15755114 * 10000; EvalErrorPrediction = 0.07370000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=0.918193s
|
||||
08/16/2016 03:04:18: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18297285 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0231s; samplesPerSecond = 10833.8
|
||||
08/16/2016 03:04:18: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12934721 * 250; EvalClassificationError = 0.04800000 * 250; time = 0.0230s; samplesPerSecond = 10872.4
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17702411 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0225s; samplesPerSecond = 11110.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14030841 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0228s; samplesPerSecond = 10941.4
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16429517 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0223s; samplesPerSecond = 11187.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19154443 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0229s; samplesPerSecond = 10924.7
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12275391 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0244s; samplesPerSecond = 10245.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16801855 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0228s; samplesPerSecond = 10987.6
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12472571 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0230s; samplesPerSecond = 10877.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19939526 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0229s; samplesPerSecond = 10895.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14222791 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0227s; samplesPerSecond = 10995.3
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12374048 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0228s; samplesPerSecond = 10962.0
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16442969 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0224s; samplesPerSecond = 11142.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19837036 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0227s; samplesPerSecond = 11003.0
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.17180200 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0235s; samplesPerSecond = 10638.3
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13326343 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0229s; samplesPerSecond = 10936.6
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14289917 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0233s; samplesPerSecond = 10727.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20692944 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0231s; samplesPerSecond = 10827.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19077197 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0231s; samplesPerSecond = 10817.4
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.14746069 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0227s; samplesPerSecond = 11027.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15464526 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0227s; samplesPerSecond = 11007.9
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13673071 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0229s; samplesPerSecond = 10923.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17348853 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0226s; samplesPerSecond = 11065.4
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14420581 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0233s; samplesPerSecond = 10725.5
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13774097 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0228s; samplesPerSecond = 10975.0
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14177905 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0225s; samplesPerSecond = 11100.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16864648 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0229s; samplesPerSecond = 10928.0
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18513623 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0228s; samplesPerSecond = 10968.3
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16393555 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0226s; samplesPerSecond = 11067.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15467676 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0228s; samplesPerSecond = 10969.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18951318 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0223s; samplesPerSecond = 11207.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13329639 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0232s; samplesPerSecond = 10793.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14604785 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0229s; samplesPerSecond = 10894.2
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13938086 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0229s; samplesPerSecond = 10896.6
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.19969873 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0226s; samplesPerSecond = 11040.9
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12584180 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0228s; samplesPerSecond = 10949.1
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18373438 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0232s; samplesPerSecond = 10780.0
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15064795 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0234s; samplesPerSecond = 10683.8
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11991260 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0229s; samplesPerSecond = 10911.3
|
||||
08/16/2016 03:04:19: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.13070557 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0230s; samplesPerSecond = 10857.8
|
||||
08/16/2016 03:04:19: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15755114 * 10000; EvalClassificationError = 0.07370000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=0.918193s
|
||||
08/16/2016 03:04:19: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030038.674053\Examples\Other\Simple2d_Simple@release_cpu/Models/simple.dnn'
|
||||
08/16/2016 03:04:19: CNTKCommandTrainEnd: Simple_Demo_Train
|
||||
|
||||
|
|
@ -636,7 +636,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -665,7 +665,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *1] -> [2 x 1 x *1]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *1], [2 x 1] -> [2 x 1 x *1]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *1] -> [2 x 1 x *1]
|
||||
Validating --> Prior = Mean (labels) : [2 x *1] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -693,8 +693,8 @@ Memory Sharing: Out of 25 matrices, 2 are shared as 1, and 23 are not shared.
|
|||
ScaledLogLikelihood : [2 x 1 x *1] }
|
||||
|
||||
BlockRandomizer::StartEpoch: epoch 0: frames [0..603] (first sequence at sample 0), data subset 0 of 1
|
||||
08/16/2016 03:04:19: Minibatch[1-1]: EvalErrorPrediction = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10734609 * 603
|
||||
08/16/2016 03:04:19: Final Results: Minibatch[1-1]: EvalErrorPrediction = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10734609 * 603; perplexity = 1.11331949
|
||||
08/16/2016 03:04:19: Minibatch[1-1]: EvalClassificationError = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10734609 * 603
|
||||
08/16/2016 03:04:19: Final Results: Minibatch[1-1]: EvalClassificationError = 0.05140962 * 603; CrossEntropyWithSoftmax = 0.10734609 * 603; perplexity = 1.11331949
|
||||
|
||||
08/16/2016 03:04:19: Action "test" complete.
|
||||
|
||||
|
|
@ -710,7 +710,7 @@ Post-processing network...
|
|||
|
||||
8 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -740,7 +740,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *2] -> [2 x 1 x *2]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *2], [2 x 1] -> [2 x 1 x *2]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *2] -> [2 x 1 x *2]
|
||||
Validating --> Prior = Mean (labels) : [2 x *2] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -764,7 +764,7 @@ Allocating matrices for forward and/or backward propagation.
|
|||
Memory Sharing: Out of 25 matrices, 3 are shared as 1, and 22 are not shared.
|
||||
|
||||
{ CrossEntropyWithSoftmax : [1]
|
||||
EvalErrorPrediction : [1]
|
||||
EvalClassificationError : [1]
|
||||
ScaledLogLikelihood : [2 x 1 x *2] }
|
||||
|
||||
Minibatch[0]: ActualMBSize = 603
|
||||
|
|
|
|||
|
|
@ -71,7 +71,7 @@ Simple_Demo_Train = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -170,7 +170,7 @@ Simple_Demo_Train = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -313,7 +313,7 @@ configparameters: Simple.cntk:Simple_Demo_Train=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 2:50*2:2
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
initValueScale = 1.0
|
||||
applyMeanVarNorm = true
|
||||
|
|
@ -381,7 +381,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -410,7 +410,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *] -> [2 x 1 x *]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *], [2 x 1] -> [2 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *], [2 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *] -> [2 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [2 x *] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -433,7 +433,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:04:26: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:04:26: Evaluation criterion node(s):
|
||||
08/16/2016 03:04:26: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 03:04:26: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -485,141 +485,141 @@ BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sampl
|
|||
BlockRandomizer::StartEpoch: epoch 0: frames [0..10000] (first sequence at sample 0), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:04:27: Starting minibatch loop.
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70124231 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0194s; samplesPerSecond = 12887.9
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76372424 * 250; EvalErrorPrediction = 0.46400000 * 250; time = 0.0179s; samplesPerSecond = 13952.5
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72703027 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0181s; samplesPerSecond = 13829.0
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73895923 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0178s; samplesPerSecond = 14035.5
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70621924 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0178s; samplesPerSecond = 14078.2
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74767041 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0177s; samplesPerSecond = 14152.3
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75094434 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0181s; samplesPerSecond = 13803.8
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78058936 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0176s; samplesPerSecond = 14213.4
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70407129 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0180s; samplesPerSecond = 13910.5
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69555762 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0178s; samplesPerSecond = 14074.2
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70626123 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0178s; samplesPerSecond = 14061.5
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74540430 * 250; EvalErrorPrediction = 0.50800000 * 250; time = 0.0178s; samplesPerSecond = 14030.8
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70824414 * 250; EvalErrorPrediction = 0.47600000 * 250; time = 0.0179s; samplesPerSecond = 14003.2
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69895020 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0179s; samplesPerSecond = 13995.4
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70353223 * 250; EvalErrorPrediction = 0.52400000 * 250; time = 0.0176s; samplesPerSecond = 14198.1
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69346387 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0177s; samplesPerSecond = 14153.9
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74449902 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0177s; samplesPerSecond = 14157.1
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73767969 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0176s; samplesPerSecond = 14175.6
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71876855 * 250; EvalErrorPrediction = 0.48400000 * 250; time = 0.0179s; samplesPerSecond = 13987.6
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71509473 * 250; EvalErrorPrediction = 0.50400000 * 250; time = 0.0180s; samplesPerSecond = 13914.4
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69956152 * 250; EvalErrorPrediction = 0.52000000 * 250; time = 0.0179s; samplesPerSecond = 13953.2
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0179s; samplesPerSecond = 13960.2
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70736035 * 250; EvalErrorPrediction = 0.54400000 * 250; time = 0.0177s; samplesPerSecond = 14094.8
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69820508 * 250; EvalErrorPrediction = 0.56800000 * 250; time = 0.0176s; samplesPerSecond = 14205.4
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69537109 * 250; EvalErrorPrediction = 0.49600000 * 250; time = 0.0178s; samplesPerSecond = 14067.1
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69347266 * 250; EvalErrorPrediction = 0.51200000 * 250; time = 0.0179s; samplesPerSecond = 13982.1
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70801172 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0178s; samplesPerSecond = 14023.7
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69131641 * 250; EvalErrorPrediction = 0.46800000 * 250; time = 0.0177s; samplesPerSecond = 14152.3
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70370312 * 250; EvalErrorPrediction = 0.52800000 * 250; time = 0.0178s; samplesPerSecond = 14023.7
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71200195 * 250; EvalErrorPrediction = 0.43600000 * 250; time = 0.0177s; samplesPerSecond = 14133.1
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69506836 * 250; EvalErrorPrediction = 0.45200000 * 250; time = 0.0178s; samplesPerSecond = 14056.0
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69935352 * 250; EvalErrorPrediction = 0.51600000 * 250; time = 0.0179s; samplesPerSecond = 13976.6
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69887109 * 250; EvalErrorPrediction = 0.50000000 * 250; time = 0.0178s; samplesPerSecond = 14018.2
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69604492 * 250; EvalErrorPrediction = 0.49200000 * 250; time = 0.0180s; samplesPerSecond = 13874.2
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.69011719 * 250; EvalErrorPrediction = 0.48800000 * 250; time = 0.0181s; samplesPerSecond = 13820.6
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.68419531 * 250; EvalErrorPrediction = 0.46000000 * 250; time = 0.0181s; samplesPerSecond = 13831.3
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.67551367 * 250; EvalErrorPrediction = 0.32400000 * 250; time = 0.0177s; samplesPerSecond = 14140.3
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.67028516 * 250; EvalErrorPrediction = 0.40000000 * 250; time = 0.0180s; samplesPerSecond = 13868.1
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.65152734 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0179s; samplesPerSecond = 13937.7
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.63594727 * 250; EvalErrorPrediction = 0.22000000 * 250; time = 0.0178s; samplesPerSecond = 14028.4
|
||||
08/16/2016 03:04:28: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70729233 * 10000; EvalErrorPrediction = 0.47740000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.717672s
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 1- 10]: CrossEntropyWithSoftmax = 0.70124231 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0194s; samplesPerSecond = 12887.9
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 11- 20]: CrossEntropyWithSoftmax = 0.76372424 * 250; EvalClassificationError = 0.46400000 * 250; time = 0.0179s; samplesPerSecond = 13952.5
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 21- 30]: CrossEntropyWithSoftmax = 0.72703027 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0181s; samplesPerSecond = 13829.0
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 31- 40]: CrossEntropyWithSoftmax = 0.73895923 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0178s; samplesPerSecond = 14035.5
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 41- 50]: CrossEntropyWithSoftmax = 0.70621924 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0178s; samplesPerSecond = 14078.2
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 51- 60]: CrossEntropyWithSoftmax = 0.74767041 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0177s; samplesPerSecond = 14152.3
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 61- 70]: CrossEntropyWithSoftmax = 0.75094434 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0181s; samplesPerSecond = 13803.8
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 71- 80]: CrossEntropyWithSoftmax = 0.78058936 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0176s; samplesPerSecond = 14213.4
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 81- 90]: CrossEntropyWithSoftmax = 0.70407129 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0180s; samplesPerSecond = 13910.5
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 91- 100]: CrossEntropyWithSoftmax = 0.69555762 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0178s; samplesPerSecond = 14074.2
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 101- 110]: CrossEntropyWithSoftmax = 0.70626123 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0178s; samplesPerSecond = 14061.5
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 111- 120]: CrossEntropyWithSoftmax = 0.74540430 * 250; EvalClassificationError = 0.50800000 * 250; time = 0.0178s; samplesPerSecond = 14030.8
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 121- 130]: CrossEntropyWithSoftmax = 0.70824414 * 250; EvalClassificationError = 0.47600000 * 250; time = 0.0179s; samplesPerSecond = 14003.2
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 131- 140]: CrossEntropyWithSoftmax = 0.69895020 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0179s; samplesPerSecond = 13995.4
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 141- 150]: CrossEntropyWithSoftmax = 0.70353223 * 250; EvalClassificationError = 0.52400000 * 250; time = 0.0176s; samplesPerSecond = 14198.1
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 151- 160]: CrossEntropyWithSoftmax = 0.69346387 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0177s; samplesPerSecond = 14153.9
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 161- 170]: CrossEntropyWithSoftmax = 0.74449902 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0177s; samplesPerSecond = 14157.1
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 171- 180]: CrossEntropyWithSoftmax = 0.73767969 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0176s; samplesPerSecond = 14175.6
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 181- 190]: CrossEntropyWithSoftmax = 0.71876855 * 250; EvalClassificationError = 0.48400000 * 250; time = 0.0179s; samplesPerSecond = 13987.6
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 191- 200]: CrossEntropyWithSoftmax = 0.71509473 * 250; EvalClassificationError = 0.50400000 * 250; time = 0.0180s; samplesPerSecond = 13914.4
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 201- 210]: CrossEntropyWithSoftmax = 0.69956152 * 250; EvalClassificationError = 0.52000000 * 250; time = 0.0179s; samplesPerSecond = 13953.2
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 211- 220]: CrossEntropyWithSoftmax = 0.69785937 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0179s; samplesPerSecond = 13960.2
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 221- 230]: CrossEntropyWithSoftmax = 0.70736035 * 250; EvalClassificationError = 0.54400000 * 250; time = 0.0177s; samplesPerSecond = 14094.8
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 231- 240]: CrossEntropyWithSoftmax = 0.69820508 * 250; EvalClassificationError = 0.56800000 * 250; time = 0.0176s; samplesPerSecond = 14205.4
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 241- 250]: CrossEntropyWithSoftmax = 0.69537109 * 250; EvalClassificationError = 0.49600000 * 250; time = 0.0178s; samplesPerSecond = 14067.1
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 251- 260]: CrossEntropyWithSoftmax = 0.69347266 * 250; EvalClassificationError = 0.51200000 * 250; time = 0.0179s; samplesPerSecond = 13982.1
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 261- 270]: CrossEntropyWithSoftmax = 0.70801172 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0178s; samplesPerSecond = 14023.7
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 271- 280]: CrossEntropyWithSoftmax = 0.69131641 * 250; EvalClassificationError = 0.46800000 * 250; time = 0.0177s; samplesPerSecond = 14152.3
|
||||
08/16/2016 03:04:27: Epoch[ 1 of 3]-Minibatch[ 281- 290]: CrossEntropyWithSoftmax = 0.70370312 * 250; EvalClassificationError = 0.52800000 * 250; time = 0.0178s; samplesPerSecond = 14023.7
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 291- 300]: CrossEntropyWithSoftmax = 0.71200195 * 250; EvalClassificationError = 0.43600000 * 250; time = 0.0177s; samplesPerSecond = 14133.1
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 301- 310]: CrossEntropyWithSoftmax = 0.69506836 * 250; EvalClassificationError = 0.45200000 * 250; time = 0.0178s; samplesPerSecond = 14056.0
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 311- 320]: CrossEntropyWithSoftmax = 0.69935352 * 250; EvalClassificationError = 0.51600000 * 250; time = 0.0179s; samplesPerSecond = 13976.6
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 321- 330]: CrossEntropyWithSoftmax = 0.69887109 * 250; EvalClassificationError = 0.50000000 * 250; time = 0.0178s; samplesPerSecond = 14018.2
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 331- 340]: CrossEntropyWithSoftmax = 0.69604492 * 250; EvalClassificationError = 0.49200000 * 250; time = 0.0180s; samplesPerSecond = 13874.2
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 341- 350]: CrossEntropyWithSoftmax = 0.69011719 * 250; EvalClassificationError = 0.48800000 * 250; time = 0.0181s; samplesPerSecond = 13820.6
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 351- 360]: CrossEntropyWithSoftmax = 0.68419531 * 250; EvalClassificationError = 0.46000000 * 250; time = 0.0181s; samplesPerSecond = 13831.3
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 361- 370]: CrossEntropyWithSoftmax = 0.67551367 * 250; EvalClassificationError = 0.32400000 * 250; time = 0.0177s; samplesPerSecond = 14140.3
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 371- 380]: CrossEntropyWithSoftmax = 0.67028516 * 250; EvalClassificationError = 0.40000000 * 250; time = 0.0180s; samplesPerSecond = 13868.1
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 381- 390]: CrossEntropyWithSoftmax = 0.65152734 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0179s; samplesPerSecond = 13937.7
|
||||
08/16/2016 03:04:28: Epoch[ 1 of 3]-Minibatch[ 391- 400]: CrossEntropyWithSoftmax = 0.63594727 * 250; EvalClassificationError = 0.22000000 * 250; time = 0.0178s; samplesPerSecond = 14028.4
|
||||
08/16/2016 03:04:28: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 0.70729233 * 10000; EvalClassificationError = 0.47740000 * 10000; totalSamplesSeen = 10000; learningRatePerSample = 0.02; epochTime=0.717672s
|
||||
08/16/2016 03:04:28: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030038.674053\Examples\Other\Simple2d_Simple@release_gpu/Models/simple.dnn.1'
|
||||
|
||||
08/16/2016 03:04:28: Starting Epoch 2: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 1: frames [10000..20000] (first sequence at sample 10000), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:04:28: Starting minibatch loop.
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.61492108 * 250; EvalErrorPrediction = 0.26800000 * 250; time = 0.0183s; samplesPerSecond = 13687.4
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.59171271 * 250; EvalErrorPrediction = 0.28400000 * 250; time = 0.0180s; samplesPerSecond = 13905.9
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.53591638 * 250; EvalErrorPrediction = 0.20000000 * 250; time = 0.0178s; samplesPerSecond = 14045.7
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.51872742 * 250; EvalErrorPrediction = 0.14000000 * 250; time = 0.0181s; samplesPerSecond = 13821.3
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.48384375 * 250; EvalErrorPrediction = 0.12400000 * 250; time = 0.0177s; samplesPerSecond = 14094.0
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.43163501 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0181s; samplesPerSecond = 13790.8
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.38970386 * 250; EvalErrorPrediction = 0.12400000 * 250; time = 0.0180s; samplesPerSecond = 13915.9
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.33681616 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0180s; samplesPerSecond = 13862.7
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.31352393 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0178s; samplesPerSecond = 14010.3
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.26829492 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0179s; samplesPerSecond = 13966.5
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.24240820 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0177s; samplesPerSecond = 14094.0
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.21015820 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0179s; samplesPerSecond = 13976.6
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.22358789 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0177s; samplesPerSecond = 14153.1
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20496631 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0181s; samplesPerSecond = 13776.4
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.20070508 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0175s; samplesPerSecond = 14307.0
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.19224707 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0180s; samplesPerSecond = 13886.6
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.19326563 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0176s; samplesPerSecond = 14189.2
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21712451 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0177s; samplesPerSecond = 14109.1
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.17562354 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0177s; samplesPerSecond = 14125.9
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.18186035 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0177s; samplesPerSecond = 14111.5
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.14065234 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0179s; samplesPerSecond = 13957.9
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17710254 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0177s; samplesPerSecond = 14107.6
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.13001953 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0176s; samplesPerSecond = 14178.0
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21622949 * 250; EvalErrorPrediction = 0.10000000 * 250; time = 0.0179s; samplesPerSecond = 13949.3
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21902246 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0182s; samplesPerSecond = 13726.5
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18068799 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0179s; samplesPerSecond = 13998.5
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16232471 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0176s; samplesPerSecond = 14165.1
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13792139 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0177s; samplesPerSecond = 14102.8
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16526709 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0181s; samplesPerSecond = 13800.7
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14743457 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0177s; samplesPerSecond = 14108.4
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15089160 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0178s; samplesPerSecond = 14053.6
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12636230 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0179s; samplesPerSecond = 13932.2
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16735547 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0176s; samplesPerSecond = 14164.3
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14530957 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0178s; samplesPerSecond = 14006.4
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13859570 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0176s; samplesPerSecond = 14166.7
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14215234 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0179s; samplesPerSecond = 13940.0
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15903027 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0178s; samplesPerSecond = 14069.4
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16232520 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0177s; samplesPerSecond = 14160.3
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13596484 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0176s; samplesPerSecond = 14198.9
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15469434 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0176s; samplesPerSecond = 14185.2
|
||||
08/16/2016 03:04:28: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.24215964 * 10000; EvalErrorPrediction = 0.09440000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=0.71504s
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.61492108 * 250; EvalClassificationError = 0.26800000 * 250; time = 0.0183s; samplesPerSecond = 13687.4
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.59171271 * 250; EvalClassificationError = 0.28400000 * 250; time = 0.0180s; samplesPerSecond = 13905.9
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.53591638 * 250; EvalClassificationError = 0.20000000 * 250; time = 0.0178s; samplesPerSecond = 14045.7
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.51872742 * 250; EvalClassificationError = 0.14000000 * 250; time = 0.0181s; samplesPerSecond = 13821.3
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.48384375 * 250; EvalClassificationError = 0.12400000 * 250; time = 0.0177s; samplesPerSecond = 14094.0
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.43163501 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0181s; samplesPerSecond = 13790.8
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.38970386 * 250; EvalClassificationError = 0.12400000 * 250; time = 0.0180s; samplesPerSecond = 13915.9
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.33681616 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0180s; samplesPerSecond = 13862.7
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.31352393 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0178s; samplesPerSecond = 14010.3
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.26829492 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0179s; samplesPerSecond = 13966.5
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.24240820 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0177s; samplesPerSecond = 14094.0
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.21015820 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0179s; samplesPerSecond = 13976.6
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.22358789 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0177s; samplesPerSecond = 14153.1
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.20496631 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0181s; samplesPerSecond = 13776.4
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.20070508 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0175s; samplesPerSecond = 14307.0
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.19224707 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0180s; samplesPerSecond = 13886.6
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.19326563 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0176s; samplesPerSecond = 14189.2
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.21712451 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0177s; samplesPerSecond = 14109.1
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.17562354 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0177s; samplesPerSecond = 14125.9
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.18186035 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0177s; samplesPerSecond = 14111.5
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.14065234 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0179s; samplesPerSecond = 13957.9
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.17710254 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0177s; samplesPerSecond = 14107.6
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.13001953 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0176s; samplesPerSecond = 14178.0
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.21622949 * 250; EvalClassificationError = 0.10000000 * 250; time = 0.0179s; samplesPerSecond = 13949.3
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.21902246 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0182s; samplesPerSecond = 13726.5
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.18068799 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0179s; samplesPerSecond = 13998.5
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16232471 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0176s; samplesPerSecond = 14165.1
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.13792139 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0177s; samplesPerSecond = 14102.8
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16526709 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0181s; samplesPerSecond = 13800.7
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.14743457 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0177s; samplesPerSecond = 14108.4
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.15089160 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0178s; samplesPerSecond = 14053.6
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.12636230 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0179s; samplesPerSecond = 13932.2
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.16735547 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0176s; samplesPerSecond = 14164.3
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.14530957 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0178s; samplesPerSecond = 14006.4
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.13859570 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0176s; samplesPerSecond = 14166.7
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.14215234 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0179s; samplesPerSecond = 13940.0
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.15903027 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0178s; samplesPerSecond = 14069.4
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.16232520 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0177s; samplesPerSecond = 14160.3
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.13596484 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0176s; samplesPerSecond = 14198.9
|
||||
08/16/2016 03:04:28: Epoch[ 2 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.15469434 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0176s; samplesPerSecond = 14185.2
|
||||
08/16/2016 03:04:28: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 0.24215964 * 10000; EvalClassificationError = 0.09440000 * 10000; totalSamplesSeen = 20000; learningRatePerSample = 0.0080000004; epochTime=0.71504s
|
||||
08/16/2016 03:04:28: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030038.674053\Examples\Other\Simple2d_Simple@release_gpu/Models/simple.dnn.2'
|
||||
|
||||
08/16/2016 03:04:28: Starting Epoch 3: learning rate per sample = 0.008000 effective momentum = 0.900000 momentum as time constant = 237.3 samples
|
||||
BlockRandomizer::StartEpoch: epoch 2: frames [20000..30000] (first sequence at sample 20000), data subset 0 of 1
|
||||
|
||||
08/16/2016 03:04:28: Starting minibatch loop.
|
||||
08/16/2016 03:04:28: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18305315 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0183s; samplesPerSecond = 13632.2
|
||||
08/16/2016 03:04:28: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12945729 * 250; EvalErrorPrediction = 0.04800000 * 250; time = 0.0177s; samplesPerSecond = 14137.9
|
||||
08/16/2016 03:04:28: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17735931 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0180s; samplesPerSecond = 13886.6
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14128339 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0180s; samplesPerSecond = 13903.6
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16558209 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0176s; samplesPerSecond = 14173.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19102692 * 250; EvalErrorPrediction = 0.10800000 * 250; time = 0.0178s; samplesPerSecond = 14036.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12279083 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0181s; samplesPerSecond = 13844.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16642798 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0178s; samplesPerSecond = 14033.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12386572 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0177s; samplesPerSecond = 14110.7
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19928418 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0177s; samplesPerSecond = 14102.8
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14213635 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0179s; samplesPerSecond = 13957.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12377087 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0177s; samplesPerSecond = 14088.5
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16361621 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0178s; samplesPerSecond = 14026.0
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19886914 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0178s; samplesPerSecond = 14015.8
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.17207544 * 250; EvalErrorPrediction = 0.09200000 * 250; time = 0.0179s; samplesPerSecond = 13935.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13323437 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0180s; samplesPerSecond = 13901.2
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14397510 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0180s; samplesPerSecond = 13905.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20777515 * 250; EvalErrorPrediction = 0.10400000 * 250; time = 0.0179s; samplesPerSecond = 13964.1
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19094092 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0180s; samplesPerSecond = 13874.2
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.14731372 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0179s; samplesPerSecond = 13942.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15483569 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0177s; samplesPerSecond = 14117.1
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13625415 * 250; EvalErrorPrediction = 0.04400000 * 250; time = 0.0177s; samplesPerSecond = 14162.7
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17354004 * 250; EvalErrorPrediction = 0.08800000 * 250; time = 0.0177s; samplesPerSecond = 14094.0
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14408350 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0179s; samplesPerSecond = 13929.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13720044 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0180s; samplesPerSecond = 13895.8
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14236426 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0178s; samplesPerSecond = 14027.6
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16857861 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0179s; samplesPerSecond = 13968.8
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18606982 * 250; EvalErrorPrediction = 0.08400000 * 250; time = 0.0180s; samplesPerSecond = 13861.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16334619 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0177s; samplesPerSecond = 14094.8
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15598535 * 250; EvalErrorPrediction = 0.07200000 * 250; time = 0.0179s; samplesPerSecond = 13979.0
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18848584 * 250; EvalErrorPrediction = 0.09600000 * 250; time = 0.0178s; samplesPerSecond = 14073.4
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13281348 * 250; EvalErrorPrediction = 0.05200000 * 250; time = 0.0178s; samplesPerSecond = 14067.1
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14679150 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0178s; samplesPerSecond = 14047.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13977344 * 250; EvalErrorPrediction = 0.06800000 * 250; time = 0.0178s; samplesPerSecond = 14027.6
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20015137 * 250; EvalErrorPrediction = 0.11200000 * 250; time = 0.0181s; samplesPerSecond = 13831.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12582129 * 250; EvalErrorPrediction = 0.06000000 * 250; time = 0.0178s; samplesPerSecond = 14022.1
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18500098 * 250; EvalErrorPrediction = 0.07600000 * 250; time = 0.0180s; samplesPerSecond = 13907.4
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15147754 * 250; EvalErrorPrediction = 0.08000000 * 250; time = 0.0181s; samplesPerSecond = 13800.0
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11988379 * 250; EvalErrorPrediction = 0.05600000 * 250; time = 0.0177s; samplesPerSecond = 14089.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.13059082 * 250; EvalErrorPrediction = 0.06400000 * 250; time = 0.0181s; samplesPerSecond = 13797.7
|
||||
08/16/2016 03:04:29: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15767216 * 10000; EvalErrorPrediction = 0.07330000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=0.716967s
|
||||
08/16/2016 03:04:28: Epoch[ 3 of 3]-Minibatch[ 1- 10, 2.50%]: CrossEntropyWithSoftmax = 0.18305315 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0183s; samplesPerSecond = 13632.2
|
||||
08/16/2016 03:04:28: Epoch[ 3 of 3]-Minibatch[ 11- 20, 5.00%]: CrossEntropyWithSoftmax = 0.12945729 * 250; EvalClassificationError = 0.04800000 * 250; time = 0.0177s; samplesPerSecond = 14137.9
|
||||
08/16/2016 03:04:28: Epoch[ 3 of 3]-Minibatch[ 21- 30, 7.50%]: CrossEntropyWithSoftmax = 0.17735931 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0180s; samplesPerSecond = 13886.6
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 31- 40, 10.00%]: CrossEntropyWithSoftmax = 0.14128339 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0180s; samplesPerSecond = 13903.6
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 41- 50, 12.50%]: CrossEntropyWithSoftmax = 0.16558209 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0176s; samplesPerSecond = 14173.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 51- 60, 15.00%]: CrossEntropyWithSoftmax = 0.19102692 * 250; EvalClassificationError = 0.10800000 * 250; time = 0.0178s; samplesPerSecond = 14036.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 61- 70, 17.50%]: CrossEntropyWithSoftmax = 0.12279083 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0181s; samplesPerSecond = 13844.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 71- 80, 20.00%]: CrossEntropyWithSoftmax = 0.16642798 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0178s; samplesPerSecond = 14033.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 81- 90, 22.50%]: CrossEntropyWithSoftmax = 0.12386572 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0177s; samplesPerSecond = 14110.7
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 91- 100, 25.00%]: CrossEntropyWithSoftmax = 0.19928418 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0177s; samplesPerSecond = 14102.8
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 101- 110, 27.50%]: CrossEntropyWithSoftmax = 0.14213635 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0179s; samplesPerSecond = 13957.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 111- 120, 30.00%]: CrossEntropyWithSoftmax = 0.12377087 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0177s; samplesPerSecond = 14088.5
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 121- 130, 32.50%]: CrossEntropyWithSoftmax = 0.16361621 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0178s; samplesPerSecond = 14026.0
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 131- 140, 35.00%]: CrossEntropyWithSoftmax = 0.19886914 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0178s; samplesPerSecond = 14015.8
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 141- 150, 37.50%]: CrossEntropyWithSoftmax = 0.17207544 * 250; EvalClassificationError = 0.09200000 * 250; time = 0.0179s; samplesPerSecond = 13935.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 151- 160, 40.00%]: CrossEntropyWithSoftmax = 0.13323437 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0180s; samplesPerSecond = 13901.2
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 161- 170, 42.50%]: CrossEntropyWithSoftmax = 0.14397510 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0180s; samplesPerSecond = 13905.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 171- 180, 45.00%]: CrossEntropyWithSoftmax = 0.20777515 * 250; EvalClassificationError = 0.10400000 * 250; time = 0.0179s; samplesPerSecond = 13964.1
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 181- 190, 47.50%]: CrossEntropyWithSoftmax = 0.19094092 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0180s; samplesPerSecond = 13874.2
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 191- 200, 50.00%]: CrossEntropyWithSoftmax = 0.14731372 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0179s; samplesPerSecond = 13942.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 201- 210, 52.50%]: CrossEntropyWithSoftmax = 0.15483569 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0177s; samplesPerSecond = 14117.1
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 211- 220, 55.00%]: CrossEntropyWithSoftmax = 0.13625415 * 250; EvalClassificationError = 0.04400000 * 250; time = 0.0177s; samplesPerSecond = 14162.7
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 221- 230, 57.50%]: CrossEntropyWithSoftmax = 0.17354004 * 250; EvalClassificationError = 0.08800000 * 250; time = 0.0177s; samplesPerSecond = 14094.0
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 231- 240, 60.00%]: CrossEntropyWithSoftmax = 0.14408350 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0179s; samplesPerSecond = 13929.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 241- 250, 62.50%]: CrossEntropyWithSoftmax = 0.13720044 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0180s; samplesPerSecond = 13895.8
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 251- 260, 65.00%]: CrossEntropyWithSoftmax = 0.14236426 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0178s; samplesPerSecond = 14027.6
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 261- 270, 67.50%]: CrossEntropyWithSoftmax = 0.16857861 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0179s; samplesPerSecond = 13968.8
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 271- 280, 70.00%]: CrossEntropyWithSoftmax = 0.18606982 * 250; EvalClassificationError = 0.08400000 * 250; time = 0.0180s; samplesPerSecond = 13861.9
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 281- 290, 72.50%]: CrossEntropyWithSoftmax = 0.16334619 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0177s; samplesPerSecond = 14094.8
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 291- 300, 75.00%]: CrossEntropyWithSoftmax = 0.15598535 * 250; EvalClassificationError = 0.07200000 * 250; time = 0.0179s; samplesPerSecond = 13979.0
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 301- 310, 77.50%]: CrossEntropyWithSoftmax = 0.18848584 * 250; EvalClassificationError = 0.09600000 * 250; time = 0.0178s; samplesPerSecond = 14073.4
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 311- 320, 80.00%]: CrossEntropyWithSoftmax = 0.13281348 * 250; EvalClassificationError = 0.05200000 * 250; time = 0.0178s; samplesPerSecond = 14067.1
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 321- 330, 82.50%]: CrossEntropyWithSoftmax = 0.14679150 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0178s; samplesPerSecond = 14047.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 331- 340, 85.00%]: CrossEntropyWithSoftmax = 0.13977344 * 250; EvalClassificationError = 0.06800000 * 250; time = 0.0178s; samplesPerSecond = 14027.6
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 341- 350, 87.50%]: CrossEntropyWithSoftmax = 0.20015137 * 250; EvalClassificationError = 0.11200000 * 250; time = 0.0181s; samplesPerSecond = 13831.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 351- 360, 90.00%]: CrossEntropyWithSoftmax = 0.12582129 * 250; EvalClassificationError = 0.06000000 * 250; time = 0.0178s; samplesPerSecond = 14022.1
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 361- 370, 92.50%]: CrossEntropyWithSoftmax = 0.18500098 * 250; EvalClassificationError = 0.07600000 * 250; time = 0.0180s; samplesPerSecond = 13907.4
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 371- 380, 95.00%]: CrossEntropyWithSoftmax = 0.15147754 * 250; EvalClassificationError = 0.08000000 * 250; time = 0.0181s; samplesPerSecond = 13800.0
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 381- 390, 97.50%]: CrossEntropyWithSoftmax = 0.11988379 * 250; EvalClassificationError = 0.05600000 * 250; time = 0.0177s; samplesPerSecond = 14089.3
|
||||
08/16/2016 03:04:29: Epoch[ 3 of 3]-Minibatch[ 391- 400, 100.00%]: CrossEntropyWithSoftmax = 0.13059082 * 250; EvalClassificationError = 0.06400000 * 250; time = 0.0181s; samplesPerSecond = 13797.7
|
||||
08/16/2016 03:04:29: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 0.15767216 * 10000; EvalClassificationError = 0.07330000 * 10000; totalSamplesSeen = 30000; learningRatePerSample = 0.0080000004; epochTime=0.716967s
|
||||
08/16/2016 03:04:29: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030038.674053\Examples\Other\Simple2d_Simple@release_gpu/Models/simple.dnn'
|
||||
08/16/2016 03:04:29: CNTKCommandTrainEnd: Simple_Demo_Train
|
||||
|
||||
|
|
@ -637,7 +637,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -666,7 +666,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *1] -> [2 x 1 x *1]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *1], [2 x 1] -> [2 x 1 x *1]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *1], [2 x 1 x *1] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *1] -> [2 x 1 x *1]
|
||||
Validating --> Prior = Mean (labels) : [2 x *1] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -694,8 +694,8 @@ Memory Sharing: Out of 25 matrices, 2 are shared as 1, and 23 are not shared.
|
|||
ScaledLogLikelihood : [2 x 1 x *1] }
|
||||
|
||||
BlockRandomizer::StartEpoch: epoch 0: frames [0..603] (first sequence at sample 0), data subset 0 of 1
|
||||
08/16/2016 03:04:29: Minibatch[1-1]: EvalErrorPrediction = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10845041 * 603
|
||||
08/16/2016 03:04:29: Final Results: Minibatch[1-1]: EvalErrorPrediction = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10845041 * 603; perplexity = 1.11454964
|
||||
08/16/2016 03:04:29: Minibatch[1-1]: EvalClassificationError = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10845041 * 603
|
||||
08/16/2016 03:04:29: Final Results: Minibatch[1-1]: EvalClassificationError = 0.04975124 * 603; CrossEntropyWithSoftmax = 0.10845041 * 603; perplexity = 1.11454964
|
||||
|
||||
08/16/2016 03:04:29: Action "test" complete.
|
||||
|
||||
|
|
@ -711,7 +711,7 @@ Post-processing network...
|
|||
|
||||
8 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -741,7 +741,7 @@ Validating --> W2*H1 = Times (W2, H2) : [2 x 50], [50 x 1 x *2] -> [2 x 1 x *2]
|
|||
Validating --> B2 = LearnableParameter() : -> [2 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [2 x 1 x *2], [2 x 1] -> [2 x 1 x *2]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [2 x *2], [2 x 1 x *2] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [2 x 1 x *2] -> [2 x 1 x *2]
|
||||
Validating --> Prior = Mean (labels) : [2 x *2] -> [2]
|
||||
Validating --> LogOfPrior = Log (Prior) : [2] -> [2]
|
||||
|
|
@ -765,7 +765,7 @@ Allocating matrices for forward and/or backward propagation.
|
|||
Memory Sharing: Out of 25 matrices, 3 are shared as 1, and 22 are not shared.
|
||||
|
||||
{ CrossEntropyWithSoftmax : [1]
|
||||
EvalErrorPrediction : [1]
|
||||
EvalClassificationError : [1]
|
||||
ScaledLogLikelihood : [2 x 1 x *2] }
|
||||
|
||||
Minibatch[0]: ActualMBSize = 603
|
||||
|
|
|
|||
|
|
@ -80,7 +80,7 @@ speechTrain = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -155,7 +155,7 @@ speechTrain = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -231,7 +231,7 @@ configparameters: FeedForward.cntk:speechTrain=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -319,7 +319,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -348,7 +348,7 @@ Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -> [132 x 1 x
|
|||
Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -371,7 +371,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:01:42: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:01:42: Evaluation criterion node(s):
|
||||
08/16/2016 10:01:42: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 10:01:42: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -424,7 +424,7 @@ requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 1
|
|||
minibatchiterator: epoch 0: frames [0..2048] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
|
||||
08/16/2016 10:01:43: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
08/16/2016 10:01:44: Finished Epoch[ 1 of 1]: [Training] CrossEntropyWithSoftmax = 4.45117986 * 2048; EvalErrorPrediction = 0.92187500 * 2048; totalSamplesSeen = 2048; learningRatePerSample = 0.00390625; epochTime=0.209966s
|
||||
08/16/2016 10:01:44: Finished Epoch[ 1 of 1]: [Training] CrossEntropyWithSoftmax = 4.45117986 * 2048; EvalClassificationError = 0.92187500 * 2048; totalSamplesSeen = 2048; learningRatePerSample = 0.00390625; epochTime=0.209966s
|
||||
08/16/2016 10:01:44: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100054.995555/Examples/Speech/AN4_FeedForward@release_cpu/Models/cntkSpeechFF.dnn'
|
||||
08/16/2016 10:01:44: CNTKCommandTrainEnd: speechTrain
|
||||
|
||||
|
|
|
|||
|
|
@ -80,7 +80,7 @@ speechTrain = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -155,7 +155,7 @@ speechTrain = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -231,7 +231,7 @@ configparameters: FeedForward.cntk:speechTrain=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -320,7 +320,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -349,7 +349,7 @@ Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -> [132 x 1 x
|
|||
Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -372,7 +372,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:01:46: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:01:46: Evaluation criterion node(s):
|
||||
08/16/2016 10:01:46: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 10:01:46: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -425,7 +425,7 @@ requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 1
|
|||
minibatchiterator: epoch 0: frames [0..2048] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
|
||||
08/16/2016 10:01:46: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
08/16/2016 10:01:46: Finished Epoch[ 1 of 1]: [Training] CrossEntropyWithSoftmax = 4.41144794 * 2048; EvalErrorPrediction = 0.92773438 * 2048; totalSamplesSeen = 2048; learningRatePerSample = 0.00390625; epochTime=0.023072s
|
||||
08/16/2016 10:01:46: Finished Epoch[ 1 of 1]: [Training] CrossEntropyWithSoftmax = 4.41144794 * 2048; EvalClassificationError = 0.92773438 * 2048; totalSamplesSeen = 2048; learningRatePerSample = 0.00390625; epochTime=0.023072s
|
||||
08/16/2016 10:01:46: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100054.995555/Examples/Speech/AN4_FeedForward@release_gpu/Models/cntkSpeechFF.dnn'
|
||||
08/16/2016 10:01:46: CNTKCommandTrainEnd: speechTrain
|
||||
|
||||
|
|
|
|||
|
|
@ -79,7 +79,7 @@ speechTrain = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -154,7 +154,7 @@ speechTrain = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -230,7 +230,7 @@ configparameters: FeedForward.cntk:speechTrain=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -318,7 +318,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -347,7 +347,7 @@ Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -> [132 x 1 x
|
|||
Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -370,7 +370,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:20:12: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:20:12: Evaluation criterion node(s):
|
||||
08/16/2016 03:20:12: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 03:20:12: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -423,7 +423,7 @@ requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 1
|
|||
minibatchiterator: epoch 0: frames [0..2048] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
|
||||
08/16/2016 03:20:15: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
08/16/2016 03:20:15: Finished Epoch[ 1 of 1]: [Training] CrossEntropyWithSoftmax = 4.46427900 * 2048; EvalErrorPrediction = 0.91259766 * 2048; totalSamplesSeen = 2048; learningRatePerSample = 0.00390625; epochTime=0.28059s
|
||||
08/16/2016 03:20:15: Finished Epoch[ 1 of 1]: [Training] CrossEntropyWithSoftmax = 4.46427900 * 2048; EvalClassificationError = 0.91259766 * 2048; totalSamplesSeen = 2048; learningRatePerSample = 0.00390625; epochTime=0.28059s
|
||||
08/16/2016 03:20:15: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031849.416502\Examples\Speech\AN4_FeedForward@release_cpu/Models/cntkSpeechFF.dnn'
|
||||
08/16/2016 03:20:15: CNTKCommandTrainEnd: speechTrain
|
||||
|
||||
|
|
|
|||
|
|
@ -79,7 +79,7 @@ speechTrain = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -154,7 +154,7 @@ speechTrain = [
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -230,7 +230,7 @@ configparameters: FeedForward.cntk:speechTrain=[
|
|||
SimpleNetworkBuilder = [
|
||||
layerSizes = 363:512:512:132
|
||||
trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
evalCriterion = "ErrorPrediction"
|
||||
evalCriterion = "ClassificationError"
|
||||
layerTypes = "Sigmoid"
|
||||
applyMeanVarNorm = true
|
||||
needPrior = true
|
||||
|
|
@ -319,7 +319,7 @@ Post-processing network...
|
|||
|
||||
7 roots:
|
||||
CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
EvalErrorPrediction = ErrorPrediction()
|
||||
EvalClassificationError = ClassificationError()
|
||||
InvStdOfFeatures = InvStdDev()
|
||||
MeanOfFeatures = Mean()
|
||||
PosteriorProb = Softmax()
|
||||
|
|
@ -348,7 +348,7 @@ Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -> [132 x 1 x
|
|||
Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -371,7 +371,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:20:20: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:20:20: Evaluation criterion node(s):
|
||||
08/16/2016 03:20:20: EvalErrorPrediction = ErrorPrediction
|
||||
08/16/2016 03:20:20: EvalClassificationError = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -424,7 +424,7 @@ requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 1
|
|||
minibatchiterator: epoch 0: frames [0..2048] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
|
||||
08/16/2016 03:20:21: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 1, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
08/16/2016 03:20:21: Finished Epoch[ 1 of 1]: [Training] CrossEntropyWithSoftmax = 4.41144794 * 2048; EvalErrorPrediction = 0.92773438 * 2048; totalSamplesSeen = 2048; learningRatePerSample = 0.00390625; epochTime=0.05551s
|
||||
08/16/2016 03:20:21: Finished Epoch[ 1 of 1]: [Training] CrossEntropyWithSoftmax = 4.41144794 * 2048; EvalClassificationError = 0.92773438 * 2048; totalSamplesSeen = 2048; learningRatePerSample = 0.00390625; epochTime=0.05551s
|
||||
08/16/2016 03:20:21: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031849.416502\Examples\Speech\AN4_FeedForward@release_gpu/Models/cntkSpeechFF.dnn'
|
||||
08/16/2016 03:20:21: CNTKCommandTrainEnd: speechTrain
|
||||
|
||||
|
|
|
|||
|
|
@ -297,7 +297,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
logPrior.prior = Mean()
|
||||
|
|
@ -455,7 +455,7 @@ Validating --> unnamed193 = Times (W, LSTMoutput3.output) : [132 x 512 x 1], [51
|
|||
Validating --> b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> LSTMoutputW = Plus (unnamed193, b) : [132 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> logPrior.prior = Mean (labels) : [132 x *] -> [132]
|
||||
Validating --> logPrior.logPrior = Log (logPrior.prior) : [132] -> [132]
|
||||
Validating --> scaledLogLikelihood = Minus (LSTMoutputW, logPrior.logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -495,7 +495,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:01:48: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:01:48: Evaluation criterion node(s):
|
||||
08/16/2016 10:01:48: err = ErrorPrediction
|
||||
08/16/2016 10:01:48: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -298,7 +298,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
logPrior.prior = Mean()
|
||||
|
|
@ -456,7 +456,7 @@ Validating --> unnamed193 = Times (W, LSTMoutput3.output) : [132 x 512 x 1], [51
|
|||
Validating --> b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> LSTMoutputW = Plus (unnamed193, b) : [132 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> logPrior.prior = Mean (labels) : [132 x *] -> [132]
|
||||
Validating --> logPrior.logPrior = Log (logPrior.prior) : [132] -> [132]
|
||||
Validating --> scaledLogLikelihood = Minus (LSTMoutputW, logPrior.logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -496,7 +496,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:02:01: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:02:01: Evaluation criterion node(s):
|
||||
08/16/2016 10:02:01: err = ErrorPrediction
|
||||
08/16/2016 10:02:01: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -296,7 +296,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
logPrior.prior = Mean()
|
||||
|
|
@ -454,7 +454,7 @@ Validating --> unnamed193 = Times (W, LSTMoutput3.output) : [132 x 512 x 1], [51
|
|||
Validating --> b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> LSTMoutputW = Plus (unnamed193, b) : [132 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> logPrior.prior = Mean (labels) : [132 x *] -> [132]
|
||||
Validating --> logPrior.logPrior = Log (logPrior.prior) : [132] -> [132]
|
||||
Validating --> scaledLogLikelihood = Minus (LSTMoutputW, logPrior.logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -494,7 +494,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:20:24: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:20:24: Evaluation criterion node(s):
|
||||
08/16/2016 03:20:24: err = ErrorPrediction
|
||||
08/16/2016 03:20:24: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -297,7 +297,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
logPrior.prior = Mean()
|
||||
|
|
@ -455,7 +455,7 @@ Validating --> unnamed193 = Times (W, LSTMoutput3.output) : [132 x 512 x 1], [51
|
|||
Validating --> b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> LSTMoutputW = Plus (unnamed193, b) : [132 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> logPrior.prior = Mean (labels) : [132 x *] -> [132]
|
||||
Validating --> logPrior.logPrior = Log (logPrior.prior) : [132] -> [132]
|
||||
Validating --> scaledLogLikelihood = Minus (LSTMoutputW, logPrior.logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -495,7 +495,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:20:44: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:20:44: Evaluation criterion node(s):
|
||||
08/16/2016 03:20:44: err = ErrorPrediction
|
||||
08/16/2016 03:20:44: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -241,7 +241,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
logPrior.prior = Mean()
|
||||
|
|
@ -385,7 +385,7 @@ Validating --> unnamed174 = Times (W, LSTMoutput3.output) : [132 x 512], [512 x
|
|||
Validating --> b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> LSTMoutputW = Plus (unnamed174, b) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> logPrior.prior = Mean (labels) : [132 x *] -> [132]
|
||||
Validating --> logPrior.logPrior = Log (logPrior.prior) : [132] -> [132]
|
||||
Validating --> scaledLogLikelihood = Minus (LSTMoutputW, logPrior.logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -426,7 +426,7 @@ Post-processing network complete.
|
|||
|
||||
05/03/2016 13:22:29: Evaluation criterion node(s):
|
||||
|
||||
05/03/2016 13:22:29: err = ErrorPrediction
|
||||
05/03/2016 13:22:29: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -242,7 +242,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
logPrior.prior = Mean()
|
||||
|
|
@ -386,7 +386,7 @@ Validating --> unnamed174 = Times (W, LSTMoutput3.output) : [132 x 512], [512 x
|
|||
Validating --> b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> LSTMoutputW = Plus (unnamed174, b) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, LSTMoutputW) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> logPrior.prior = Mean (labels) : [132 x *] -> [132]
|
||||
Validating --> logPrior.logPrior = Log (logPrior.prior) : [132] -> [132]
|
||||
Validating --> scaledLogLikelihood = Minus (LSTMoutputW, logPrior.logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -427,7 +427,7 @@ Post-processing network complete.
|
|||
|
||||
05/03/2016 13:22:43: Evaluation criterion node(s):
|
||||
|
||||
05/03/2016 13:22:43: err = ErrorPrediction
|
||||
05/03/2016 13:22:43: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
|
|
@ -376,7 +376,7 @@ Node 'W1' (LearnableParameter operation): Initializing Parameter[183 x 1024] <-
|
|||
Post-processing network...
|
||||
|
||||
6 roots:
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
cr = CrossEntropyWithSoftmax()
|
||||
featNorm.xMean = Mean()
|
||||
|
|
@ -452,7 +452,7 @@ Validating --> LSTMoutput.ot = Sigmoid (LSTMoutput.unnamed106) : [1024 x *] -> [
|
|||
Validating --> LSTMoutput.unnamed109 = Tanh (LSTMoutput.ct) : [1024 x *] -> [1024 x *]
|
||||
Validating --> LSTMoutput.output = ElementTimes (LSTMoutput.ot, LSTMoutput.unnamed109) : [1024 x *], [1024 x *] -> [1024 x *]
|
||||
Validating --> LSTMoutputW1 = Times (W1, LSTMoutput.output) : [183 x 1024], [1024 x *] -> [183 x *]
|
||||
Validating --> Err = ErrorPrediction (labels, LSTMoutputW1) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, LSTMoutputW1) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (LSTMoutputW1, logPrior.LogPrior) : [183 x *], [183] -> [183 x *]
|
||||
|
|
@ -492,7 +492,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:21:37: cr = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:21:37: Evaluation criterion node(s):
|
||||
08/16/2016 10:21:37: Err = ErrorPrediction
|
||||
08/16/2016 10:21:37: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -325,7 +325,7 @@ Microsoft::MSR::CNTK::GPUMatrix<ElemType>::SetUniformRandomValue (GPU): creating
|
|||
Post-processing network...
|
||||
|
||||
6 roots:
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
cr = CrossEntropyWithSoftmax()
|
||||
featNorm.xMean = Mean()
|
||||
|
|
@ -401,7 +401,7 @@ Validating --> LSTMoutput.ot = Sigmoid (LSTMoutput.unnamed106) : [1024 x *] -> [
|
|||
Validating --> LSTMoutput.unnamed109 = Tanh (LSTMoutput.ct) : [1024 x *] -> [1024 x *]
|
||||
Validating --> LSTMoutput.output = ElementTimes (LSTMoutput.ot, LSTMoutput.unnamed109) : [1024 x *], [1024 x *] -> [1024 x *]
|
||||
Validating --> LSTMoutputW1 = Times (W1, LSTMoutput.output) : [183 x 1024], [1024 x *] -> [183 x *]
|
||||
Validating --> Err = ErrorPrediction (labels, LSTMoutputW1) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, LSTMoutputW1) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (LSTMoutputW1, logPrior.LogPrior) : [183 x *], [183] -> [183 x *]
|
||||
|
|
@ -442,7 +442,7 @@ Post-processing network complete.
|
|||
|
||||
05/17/2016 12:55:27: Evaluation criterion node(s):
|
||||
|
||||
05/17/2016 12:55:27: Err = ErrorPrediction
|
||||
05/17/2016 12:55:27: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -302,7 +302,7 @@ Post-processing network...
|
|||
|
||||
8 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
FER = ErrorPrediction()
|
||||
FER = ClassificationError()
|
||||
Prior = Mean()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featInvStd1 = InvStdDev()
|
||||
|
|
@ -344,7 +344,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L3.S) : [183 x 512], [512 x *] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *], [183] -> [183 x *]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> FER = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> FER = ClassificationError (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Prior = Mean (labels) : [183 x *] -> [183]
|
||||
Validating --> LogPrior = Log (Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, LogPrior) : [183 x *], [183] -> [183 x *]
|
||||
|
|
@ -366,7 +366,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:36:07: CE.SM = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:36:07: Evaluation criterion node(s):
|
||||
08/16/2016 10:36:07: FER = ErrorPrediction
|
||||
08/16/2016 10:36:07: FER = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -267,7 +267,7 @@ Post-processing network...
|
|||
|
||||
8 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
FER = ErrorPrediction()
|
||||
FER = ClassificationError()
|
||||
Prior = Mean()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featInvStd1 = InvStdDev()
|
||||
|
|
@ -309,7 +309,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L3.S) : [183 x 512], [512 x *] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *], [183] -> [183 x *]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> FER = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> FER = ClassificationError (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Prior = Mean (labels) : [183 x *] -> [183]
|
||||
Validating --> LogPrior = Log (Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, LogPrior) : [183 x *], [183] -> [183 x *]
|
||||
|
|
@ -332,7 +332,7 @@ Post-processing network complete.
|
|||
|
||||
05/18/2016 01:01:36: Evaluation criterion node(s):
|
||||
|
||||
05/18/2016 01:01:36: FER = ErrorPrediction
|
||||
05/18/2016 01:01:36: FER = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -311,8 +311,8 @@ Post-processing network...
|
|||
9 roots:
|
||||
CE1 = CrossEntropyWithSoftmax()
|
||||
CE2 = CrossEntropyWithSoftmax()
|
||||
FER1 = ErrorPrediction()
|
||||
FER2 = ErrorPrediction()
|
||||
FER1 = ClassificationError()
|
||||
FER2 = ClassificationError()
|
||||
ObjFcn.O = Plus()
|
||||
Prior = Mean()
|
||||
ScaledLogLikelihood = Minus()
|
||||
|
|
@ -352,8 +352,8 @@ Validating --> BFF2.FF.T = Times (BFF2.W, L3.S) : [8 x 512], [512 x *] -> [8 x *
|
|||
Validating --> BFF2.B = LearnableParameter() : -> [8]
|
||||
Validating --> BFF2.FF.P = Plus (BFF2.FF.T, BFF2.B) : [8 x *], [8] -> [8 x *]
|
||||
Validating --> CE2 = CrossEntropyWithSoftmax (regions, BFF2.FF.P) : [8 x *], [8 x *] -> [1]
|
||||
Validating --> FER1 = ErrorPrediction (labels, BFF1.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> FER2 = ErrorPrediction (regions, BFF2.FF.P) : [8 x *], [8 x *] -> [1]
|
||||
Validating --> FER1 = ClassificationError (labels, BFF1.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> FER2 = ClassificationError (regions, BFF2.FF.P) : [8 x *], [8 x *] -> [1]
|
||||
Validating --> ObjFcn.A1 = LearnableParameter() : -> [1 x 1]
|
||||
Validating --> ObjFcn.T1 = Times (ObjFcn.A1, CE1) : [1 x 1], [1] -> [1]
|
||||
Validating --> ObjFcn.A2 = LearnableParameter() : -> [1 x 1]
|
||||
|
|
@ -381,9 +381,9 @@ Post-processing network complete.
|
|||
|
||||
08/17/2016 09:55:14: Evaluation criterion node(s):
|
||||
08/17/2016 09:55:14: CE1 = CrossEntropyWithSoftmax
|
||||
08/17/2016 09:55:14: FER1 = ErrorPrediction
|
||||
08/17/2016 09:55:14: FER1 = ClassificationError
|
||||
08/17/2016 09:55:14: CE2 = CrossEntropyWithSoftmax
|
||||
08/17/2016 09:55:14: FER2 = ErrorPrediction
|
||||
08/17/2016 09:55:14: FER2 = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -264,8 +264,8 @@ Post-processing network...
|
|||
9 roots:
|
||||
CE1 = CrossEntropyWithSoftmax()
|
||||
CE2 = CrossEntropyWithSoftmax()
|
||||
FER1 = ErrorPrediction()
|
||||
FER2 = ErrorPrediction()
|
||||
FER1 = ClassificationError()
|
||||
FER2 = ClassificationError()
|
||||
ObjFcn.O = Plus()
|
||||
Prior = Mean()
|
||||
ScaledLogLikelihood = Minus()
|
||||
|
|
@ -305,8 +305,8 @@ Validating --> BFF2.FF.T = Times (BFF2.W, L3.S) : [8 x 512], [512 x *] -> [8 x *
|
|||
Validating --> BFF2.B = LearnableParameter() : -> [8]
|
||||
Validating --> BFF2.FF.P = Plus (BFF2.FF.T, BFF2.B) : [8 x *], [8] -> [8 x *]
|
||||
Validating --> CE2 = CrossEntropyWithSoftmax (regions, BFF2.FF.P) : [8 x *], [8 x *] -> [1]
|
||||
Validating --> FER1 = ErrorPrediction (labels, BFF1.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> FER2 = ErrorPrediction (regions, BFF2.FF.P) : [8 x *], [8 x *] -> [1]
|
||||
Validating --> FER1 = ClassificationError (labels, BFF1.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> FER2 = ClassificationError (regions, BFF2.FF.P) : [8 x *], [8 x *] -> [1]
|
||||
Validating --> ObjFcn.A1 = LearnableParameter() : -> [1 x 1]
|
||||
Validating --> ObjFcn.T1 = Times (ObjFcn.A1, CE1) : [1 x 1], [1] -> [1]
|
||||
Validating --> ObjFcn.A2 = LearnableParameter() : -> [1 x 1]
|
||||
|
|
@ -335,9 +335,9 @@ Post-processing network complete.
|
|||
05/17/2016 13:12:47: Evaluation criterion node(s):
|
||||
|
||||
05/17/2016 13:12:47: CE1 = CrossEntropyWithSoftmax
|
||||
05/17/2016 13:12:47: FER1 = ErrorPrediction
|
||||
05/17/2016 13:12:47: FER1 = ClassificationError
|
||||
05/17/2016 13:12:47: CE2 = CrossEntropyWithSoftmax
|
||||
05/17/2016 13:12:47: FER2 = ErrorPrediction
|
||||
05/17/2016 13:12:47: FER2 = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -270,7 +270,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -303,7 +303,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L3.S) : [183 x 512], [512 x *] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *], [183] -> [183 x *]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (myLabels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (myLabels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (myLabels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (myLabels) : [183 x *] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *], [183] -> [183 x *]
|
||||
|
|
@ -325,7 +325,7 @@ Post-processing network complete.
|
|||
08/16/2016 14:25:37: CE.SM = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 14:25:37: Evaluation criterion node(s):
|
||||
08/16/2016 14:25:37: Err = ErrorPrediction
|
||||
08/16/2016 14:25:37: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -237,7 +237,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -270,7 +270,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L3.S) : [183 x 512], [512 x *] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *], [183] -> [183 x *]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (myLabels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (myLabels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (myLabels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (myLabels) : [183 x *] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *], [183] -> [183 x *]
|
||||
|
|
@ -293,7 +293,7 @@ Post-processing network complete.
|
|||
|
||||
05/17/2016 13:16:32: Evaluation criterion node(s):
|
||||
|
||||
05/17/2016 13:16:32: Err = ErrorPrediction
|
||||
05/17/2016 13:16:32: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
|
|
@ -379,7 +379,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -402,7 +402,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L1.S) : [183 x 512], [512 x *] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *], [183] -> [183 x *]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *], [183] -> [183 x *]
|
||||
|
|
@ -424,7 +424,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:46:34: CE.SM = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:46:34: Evaluation criterion node(s):
|
||||
08/16/2016 10:46:34: Err = ErrorPrediction
|
||||
08/16/2016 10:46:34: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1370,7 +1370,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -1393,7 +1393,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L1.S) : [183 x 512], [512 x *1] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *1], [183] -> [183 x *1]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *1] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *1], [183] -> [183 x *1]
|
||||
|
|
@ -1418,7 +1418,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -1446,7 +1446,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L2.S) : [183 x 512], [512 x *1] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *1], [183] -> [183 x *1]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *1] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *1], [183] -> [183 x *1]
|
||||
|
|
@ -1487,7 +1487,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -1515,7 +1515,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L2.S) : [183 x 512], [512 x *3] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *3], [183] -> [183 x *3]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *3], [183 x *3] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *3], [183 x *3] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *3], [183 x *3] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *3] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *3], [183] -> [183 x *3]
|
||||
|
|
@ -1537,7 +1537,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:46:48: CE.SM = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:46:48: Evaluation criterion node(s):
|
||||
08/16/2016 10:46:48: Err = ErrorPrediction
|
||||
08/16/2016 10:46:48: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -2484,7 +2484,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -2512,7 +2512,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L2.S) : [183 x 512], [512 x *4] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *4], [183] -> [183 x *4]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *4] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *4], [183] -> [183 x *4]
|
||||
|
|
@ -2537,7 +2537,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -2570,7 +2570,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L3.S) : [183 x 512], [512 x *4] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *4], [183] -> [183 x *4]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *4] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *4], [183] -> [183 x *4]
|
||||
|
|
@ -2611,7 +2611,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -2644,7 +2644,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L3.S) : [183 x 512], [512 x *6] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *6], [183] -> [183 x *6]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *6], [183 x *6] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *6], [183 x *6] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *6], [183 x *6] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *6] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *6], [183] -> [183 x *6]
|
||||
|
|
@ -2666,7 +2666,7 @@ Post-processing network complete.
|
|||
08/16/2016 10:47:03: CE.SM = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 10:47:03: Evaluation criterion node(s):
|
||||
08/16/2016 10:47:03: Err = ErrorPrediction
|
||||
08/16/2016 10:47:03: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -354,7 +354,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -377,7 +377,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L1.S) : [183 x 512], [512 x *] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *], [183] -> [183 x *]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *], [183 x *] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *], [183] -> [183 x *]
|
||||
|
|
@ -400,7 +400,7 @@ Post-processing network complete.
|
|||
|
||||
05/17/2016 13:22:43: Evaluation criterion node(s):
|
||||
|
||||
05/17/2016 13:22:43: Err = ErrorPrediction
|
||||
05/17/2016 13:22:43: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1351,7 +1351,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -1374,7 +1374,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L1.S) : [183 x 512], [512 x *1] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *1], [183] -> [183 x *1]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *1] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *1], [183] -> [183 x *1]
|
||||
|
|
@ -1395,7 +1395,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -1423,7 +1423,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L2.S) : [183 x 512], [512 x *1] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *1], [183] -> [183 x *1]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *1], [183 x *1] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *1] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *1], [183] -> [183 x *1]
|
||||
|
|
@ -1464,7 +1464,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -1492,7 +1492,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L2.S) : [183 x 512], [512 x *3] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *3], [183] -> [183 x *3]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *3], [183 x *3] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *3], [183 x *3] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *3], [183 x *3] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *3] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *3], [183] -> [183 x *3]
|
||||
|
|
@ -1515,7 +1515,7 @@ Post-processing network complete.
|
|||
|
||||
05/17/2016 13:23:11: Evaluation criterion node(s):
|
||||
|
||||
05/17/2016 13:23:11: Err = ErrorPrediction
|
||||
05/17/2016 13:23:11: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -2462,7 +2462,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -2490,7 +2490,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L2.S) : [183 x 512], [512 x *4] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *4], [183] -> [183 x *4]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *4] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *4], [183] -> [183 x *4]
|
||||
|
|
@ -2511,7 +2511,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -2544,7 +2544,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L3.S) : [183 x 512], [512 x *4] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *4], [183] -> [183 x *4]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *4], [183 x *4] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *4] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *4], [183] -> [183 x *4]
|
||||
|
|
@ -2585,7 +2585,7 @@ Post-processing network...
|
|||
|
||||
6 roots:
|
||||
CE.SM = CrossEntropyWithSoftmax()
|
||||
Err = ErrorPrediction()
|
||||
Err = ClassificationError()
|
||||
ScaledLogLikelihood = Minus()
|
||||
featNorm.xMean = Mean()
|
||||
featNorm.xStdDev = InvStdDev()
|
||||
|
|
@ -2618,7 +2618,7 @@ Validating --> CE.BFF.FF.T = Times (CE.BFF.W, L3.S) : [183 x 512], [512 x *6] ->
|
|||
Validating --> CE.BFF.B = LearnableParameter() : -> [183]
|
||||
Validating --> CE.BFF.FF.P = Plus (CE.BFF.FF.T, CE.BFF.B) : [183 x *6], [183] -> [183 x *6]
|
||||
Validating --> CE.SM = CrossEntropyWithSoftmax (labels, CE.BFF.FF.P) : [183 x *6], [183 x *6] -> [1]
|
||||
Validating --> Err = ErrorPrediction (labels, CE.BFF.FF.P) : [183 x *6], [183 x *6] -> [1]
|
||||
Validating --> Err = ClassificationError (labels, CE.BFF.FF.P) : [183 x *6], [183 x *6] -> [1]
|
||||
Validating --> logPrior.Prior = Mean (labels) : [183 x *6] -> [183]
|
||||
Validating --> logPrior.LogPrior = Log (logPrior.Prior) : [183] -> [183]
|
||||
Validating --> ScaledLogLikelihood = Minus (CE.BFF.FF.P, logPrior.LogPrior) : [183 x *6], [183] -> [183 x *6]
|
||||
|
|
@ -2641,7 +2641,7 @@ Post-processing network complete.
|
|||
|
||||
05/17/2016 13:23:35: Evaluation criterion node(s):
|
||||
|
||||
05/17/2016 13:23:35: Err = ErrorPrediction
|
||||
05/17/2016 13:23:35: Err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
|
|
@ -394,7 +394,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -449,7 +449,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *], [1000] -> [1000 x *]
|
||||
Validating --> labels = InputValue() : -> [1000 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -484,7 +484,7 @@ Post-processing network complete.
|
|||
08/16/2016 09:55:25: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 09:55:25: Evaluation criterion node(s):
|
||||
08/16/2016 09:55:25: err = ErrorPrediction
|
||||
08/16/2016 09:55:25: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -615,7 +615,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -666,7 +666,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *1], [1000] -> [1000 x *1]
|
||||
Validating --> labels = InputValue() : -> [1000 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -703,8 +703,8 @@ Post-processing network...
|
|||
4 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
errTop5 = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
errTop5 = ClassificationError()
|
||||
|
||||
Validating network. 50 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -755,9 +755,9 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *1], [1000] -> [1000 x *1]
|
||||
Validating --> labels = InputValue() : -> [1000 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> unnamed143 = LearnableParameter() : -> [1 x 1]
|
||||
Validating --> errTop5 = ErrorPrediction (labels, OutputNodes.z, unnamed143) : [1000 x *1], [1000 x *1], [1 x 1] -> [1]
|
||||
Validating --> errTop5 = ClassificationError (labels, OutputNodes.z, unnamed143) : [1000 x *1], [1000 x *1], [1 x 1] -> [1]
|
||||
|
||||
Validating network. 31 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -817,7 +817,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -872,7 +872,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *2], [1000] -> [1000 x *2]
|
||||
Validating --> labels = InputValue() : -> [1000 x *2]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -391,7 +391,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -446,7 +446,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *], [1000] -> [1000 x *]
|
||||
Validating --> labels = InputValue() : -> [1000 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -481,7 +481,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:03:45: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:03:45: Evaluation criterion node(s):
|
||||
08/16/2016 03:03:45: err = ErrorPrediction
|
||||
08/16/2016 03:03:45: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -612,7 +612,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -663,7 +663,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *1], [1000] -> [1000 x *1]
|
||||
Validating --> labels = InputValue() : -> [1000 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -700,8 +700,8 @@ Post-processing network...
|
|||
4 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
errTop5 = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
errTop5 = ClassificationError()
|
||||
|
||||
Validating network. 50 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -752,9 +752,9 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *1], [1000] -> [1000 x *1]
|
||||
Validating --> labels = InputValue() : -> [1000 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> unnamed143 = LearnableParameter() : -> [1 x 1]
|
||||
Validating --> errTop5 = ErrorPrediction (labels, OutputNodes.z, unnamed143) : [1000 x *1], [1000 x *1], [1 x 1] -> [1]
|
||||
Validating --> errTop5 = ClassificationError (labels, OutputNodes.z, unnamed143) : [1000 x *1], [1000 x *1], [1 x 1] -> [1]
|
||||
|
||||
Validating network. 31 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -814,7 +814,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -869,7 +869,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *2], [1000] -> [1000 x *2]
|
||||
Validating --> labels = InputValue() : -> [1000 x *2]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -394,7 +394,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -449,7 +449,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *], [1000] -> [1000 x *]
|
||||
Validating --> labels = InputValue() : -> [1000 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -484,7 +484,7 @@ Post-processing network complete.
|
|||
08/16/2016 09:55:25: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 09:55:25: Evaluation criterion node(s):
|
||||
08/16/2016 09:55:25: err = ErrorPrediction
|
||||
08/16/2016 09:55:25: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -615,7 +615,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -666,7 +666,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *1], [1000] -> [1000 x *1]
|
||||
Validating --> labels = InputValue() : -> [1000 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -703,8 +703,8 @@ Post-processing network...
|
|||
4 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
errTop5 = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
errTop5 = ClassificationError()
|
||||
|
||||
Validating network. 50 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -755,9 +755,9 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *1], [1000] -> [1000 x *1]
|
||||
Validating --> labels = InputValue() : -> [1000 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> unnamed143 = LearnableParameter() : -> [1 x 1]
|
||||
Validating --> errTop5 = ErrorPrediction (labels, OutputNodes.z, unnamed143) : [1000 x *1], [1000 x *1], [1 x 1] -> [1]
|
||||
Validating --> errTop5 = ClassificationError (labels, OutputNodes.z, unnamed143) : [1000 x *1], [1000 x *1], [1 x 1] -> [1]
|
||||
|
||||
Validating network. 31 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -817,7 +817,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -872,7 +872,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *2], [1000] -> [1000 x *2]
|
||||
Validating --> labels = InputValue() : -> [1000 x *2]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -391,7 +391,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -446,7 +446,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *], [1000] -> [1000 x *]
|
||||
Validating --> labels = InputValue() : -> [1000 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *], [1000 x *] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -481,7 +481,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:03:45: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:03:45: Evaluation criterion node(s):
|
||||
08/16/2016 03:03:45: err = ErrorPrediction
|
||||
08/16/2016 03:03:45: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -612,7 +612,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -663,7 +663,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *1], [1000] -> [1000 x *1]
|
||||
Validating --> labels = InputValue() : -> [1000 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -700,8 +700,8 @@ Post-processing network...
|
|||
4 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
errTop5 = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
errTop5 = ClassificationError()
|
||||
|
||||
Validating network. 50 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -752,9 +752,9 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *1], [1000] -> [1000 x *1]
|
||||
Validating --> labels = InputValue() : -> [1000 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *1], [1000 x *1] -> [1]
|
||||
Validating --> unnamed143 = LearnableParameter() : -> [1 x 1]
|
||||
Validating --> errTop5 = ErrorPrediction (labels, OutputNodes.z, unnamed143) : [1000 x *1], [1000 x *1], [1 x 1] -> [1]
|
||||
Validating --> errTop5 = ClassificationError (labels, OutputNodes.z, unnamed143) : [1000 x *1], [1000 x *1], [1 x 1] -> [1]
|
||||
|
||||
Validating network. 31 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -814,7 +814,7 @@ Post-processing network...
|
|||
3 roots:
|
||||
OutputNodes.z = Plus()
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 48 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -869,7 +869,7 @@ Validating --> OutputNodes.b = LearnableParameter() : -> [1000]
|
|||
Validating --> OutputNodes.z = Plus (OutputNodes.t, OutputNodes.b) : [1000 x *2], [1000] -> [1000 x *2]
|
||||
Validating --> labels = InputValue() : -> [1000 x *2]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OutputNodes.z) : [1000 x *2], [1000 x *2] -> [1]
|
||||
|
||||
Validating network. 30 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -120,7 +120,7 @@ train = [
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -239,7 +239,7 @@ train = [
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -383,7 +383,7 @@ configparameters: cntk.cntk:train=[
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -447,7 +447,7 @@ Post-processing network...
|
|||
|
||||
2 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 28 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -482,7 +482,7 @@ Validating --> ol.out.PlusArgs[0] = Times (ol.W, h1) : [10 x 128], [128 x *] ->
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol = Plus (ol.out.PlusArgs[0], ol.b) : [10 x *], [10 x 1] -> [10 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
|
||||
Validating network. 17 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -509,7 +509,7 @@ Post-processing network complete.
|
|||
08/16/2016 09:56:36: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 09:56:36: Evaluation criterion node(s):
|
||||
08/16/2016 09:56:36: err = ErrorPrediction
|
||||
08/16/2016 09:56:36: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -618,7 +618,7 @@ Post-processing network...
|
|||
|
||||
2 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 28 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -649,7 +649,7 @@ Validating --> ol.out.PlusArgs[0] = Times (ol.W, h1) : [10 x 128], [128 x *1] ->
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol = Plus (ol.out.PlusArgs[0], ol.b) : [10 x *1], [10 x 1] -> [10 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
|
||||
Validating network. 17 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -803,7 +803,7 @@ train = [
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -923,7 +923,7 @@ train = [
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -1069,7 +1069,7 @@ configparameters: cntk.cntk:train=[
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -1117,7 +1117,7 @@ Post-processing network...
|
|||
|
||||
2 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 28 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -1148,7 +1148,7 @@ Validating --> ol.out.PlusArgs[0] = Times (ol.W, h1) : [10 x 128], [128 x *] ->
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol = Plus (ol.out.PlusArgs[0], ol.b) : [10 x *], [10 x 1] -> [10 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
|
||||
Validating network. 17 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -1175,7 +1175,7 @@ Post-processing network complete.
|
|||
08/16/2016 09:56:39: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 09:56:39: Evaluation criterion node(s):
|
||||
08/16/2016 09:56:39: err = ErrorPrediction
|
||||
08/16/2016 09:56:39: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1252,7 +1252,7 @@ Post-processing network...
|
|||
|
||||
2 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 28 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -1283,7 +1283,7 @@ Validating --> ol.out.PlusArgs[0] = Times (ol.W, h1) : [10 x 128], [128 x *1] ->
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol = Plus (ol.out.PlusArgs[0], ol.b) : [10 x *1], [10 x 1] -> [10 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
|
||||
Validating network. 17 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
|
|
@ -117,7 +117,7 @@ train = [
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -236,7 +236,7 @@ train = [
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -380,7 +380,7 @@ configparameters: cntk.cntk:train=[
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -444,7 +444,7 @@ Post-processing network...
|
|||
|
||||
2 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 28 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -479,7 +479,7 @@ Validating --> ol.out.PlusArgs[0] = Times (ol.W, h1) : [10 x 128], [128 x *] ->
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol = Plus (ol.out.PlusArgs[0], ol.b) : [10 x *], [10 x 1] -> [10 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
|
||||
Validating network. 17 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -506,7 +506,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:05:48: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:05:48: Evaluation criterion node(s):
|
||||
08/16/2016 03:05:48: err = ErrorPrediction
|
||||
08/16/2016 03:05:48: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -615,7 +615,7 @@ Post-processing network...
|
|||
|
||||
2 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 28 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -646,7 +646,7 @@ Validating --> ol.out.PlusArgs[0] = Times (ol.W, h1) : [10 x 128], [128 x *1] ->
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol = Plus (ol.out.PlusArgs[0], ol.b) : [10 x *1], [10 x 1] -> [10 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
|
||||
Validating network. 17 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -797,7 +797,7 @@ train = [
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -917,7 +917,7 @@ train = [
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -1063,7 +1063,7 @@ configparameters: cntk.cntk:train=[
|
|||
h1 = DNNSigmoidLayer(4 * 4 * cMap2, h1Dim, FlattenDimensions (pool2, 1, 3), 1).out
|
||||
ol = DNNLayer(h1Dim, labelDim, h1, 1).out
|
||||
ce = CrossEntropyWithSoftmax(labels, ol, tag="criterion")
|
||||
err = ErrorPrediction(labels, ol, tag="evaluation")
|
||||
err = ClassificationError(labels, ol, tag="evaluation")
|
||||
]
|
||||
SGD = [
|
||||
epochSize = 1000
|
||||
|
|
@ -1111,7 +1111,7 @@ Post-processing network...
|
|||
|
||||
2 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 28 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -1142,7 +1142,7 @@ Validating --> ol.out.PlusArgs[0] = Times (ol.W, h1) : [10 x 128], [128 x *] ->
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol = Plus (ol.out.PlusArgs[0], ol.b) : [10 x *], [10 x 1] -> [10 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, ol) : [10 x *], [10 x 1 x *] -> [1]
|
||||
|
||||
Validating network. 17 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -1169,7 +1169,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:05:52: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:05:52: Evaluation criterion node(s):
|
||||
08/16/2016 03:05:52: err = ErrorPrediction
|
||||
08/16/2016 03:05:52: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1246,7 +1246,7 @@ Post-processing network...
|
|||
|
||||
2 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
|
||||
Validating network. 28 nodes to process in pass 1.
|
||||
|
||||
|
|
@ -1277,7 +1277,7 @@ Validating --> ol.out.PlusArgs[0] = Times (ol.W, h1) : [10 x 128], [128 x *1] ->
|
|||
Validating --> ol.b = LearnableParameter() : -> [10 x 1]
|
||||
Validating --> ol = Plus (ol.out.PlusArgs[0], ol.b) : [10 x *1], [10 x 1] -> [10 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, ol) : [10 x *1], [10 x 1 x *1] -> [1]
|
||||
|
||||
Validating network. 17 nodes to process in pass 2.
|
||||
|
||||
|
|
|
|||
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
Разница между файлами не показана из-за своего большого размера
Загрузить разницу
|
|
@ -541,7 +541,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 1]) -> [512, MBSize 1]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 1]) -> [132, MBSize 1]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 1], OL.b[132, 1]) -> [132, MBSize 1]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
|
||||
Validating for node Err. 1 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -559,7 +559,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 1]) -> [512, MBSize 1]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 1]) -> [132, MBSize 1]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 1], OL.b[132, 1]) -> [132, MBSize 1]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -577,7 +577,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 1]) -> [512, MBSize 1]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 1]) -> [132, MBSize 1]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 1], OL.b[132, 1]) -> [132, MBSize 1]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
|
||||
7 out of 15 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -599,7 +599,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 1]) -> [512, MBSize 1]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 1]) -> [132, MBSize 1]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 1], OL.b[132, 1]) -> [132, MBSize 1]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -617,7 +617,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 1]) -> [512, MBSize 1]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 1]) -> [132, MBSize 1]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 1], OL.b[132, 1]) -> [132, MBSize 1]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
|
||||
7 out of 15 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -938,7 +938,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err. 1 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -956,7 +956,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -974,7 +974,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
7 out of 15 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -996,7 +996,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -1014,7 +1014,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
7 out of 15 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -1291,7 +1291,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -1314,7 +1314,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -1341,7 +1341,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -1364,7 +1364,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -1675,7 +1675,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err. 1 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -1698,7 +1698,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -1721,7 +1721,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -1748,7 +1748,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -1771,7 +1771,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -2152,7 +2152,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err. 1 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -2175,7 +2175,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -2198,7 +2198,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -2225,7 +2225,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -2248,7 +2248,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -2580,7 +2580,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -2608,7 +2608,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
11 out of 25 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -2640,7 +2640,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -2668,7 +2668,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
11 out of 25 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -3039,7 +3039,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err. 1 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -3067,7 +3067,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -3095,7 +3095,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
11 out of 25 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -3127,7 +3127,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -3155,7 +3155,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
11 out of 25 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
|
|||
|
|
@ -395,7 +395,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 19 nodes to process in pass 1.
|
||||
|
|
@ -415,7 +415,7 @@ Validating --> OL.t = Times (OL.W, HL1.y) : [132 x 512], [512 x 1 x *] -> [132 x
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
|
|
@ -437,7 +437,7 @@ Post-processing network complete.
|
|||
08/16/2016 09:57:16: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 09:57:16: Evaluation criterion node(s):
|
||||
08/16/2016 09:57:16: err = ErrorPrediction
|
||||
08/16/2016 09:57:16: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -560,7 +560,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 19 nodes to process in pass 1.
|
||||
|
|
@ -580,7 +580,7 @@ Validating --> OL.t = Times (OL.W, HL1.y) : [132 x 512], [512 x 1 x *1] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *1], [132 x 1] -> [132 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *1], [132 x 1] -> [132 x 1 x *1]
|
||||
|
|
@ -605,7 +605,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 24 nodes to process in pass 1.
|
||||
|
|
@ -630,7 +630,7 @@ Validating --> OL.t = Times (OL.W, HL2.y) : [132 x 512], [512 x 1 x *1] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *1], [132 x 1] -> [132 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *1], [132 x 1] -> [132 x 1 x *1]
|
||||
|
|
@ -671,7 +671,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 24 nodes to process in pass 1.
|
||||
|
|
@ -696,7 +696,7 @@ Validating --> OL.t = Times (OL.W, HL2.y) : [132 x 512], [512 x 1 x *3] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *3], [132 x 1] -> [132 x 1 x *3]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *3], [132 x 1 x *3] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *3], [132 x 1 x *3] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *3], [132 x 1 x *3] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *3], [132 x 1] -> [132 x 1 x *3]
|
||||
|
|
@ -718,7 +718,7 @@ Post-processing network complete.
|
|||
08/16/2016 09:57:17: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 09:57:17: Evaluation criterion node(s):
|
||||
08/16/2016 09:57:17: err = ErrorPrediction
|
||||
08/16/2016 09:57:17: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -851,7 +851,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 24 nodes to process in pass 1.
|
||||
|
|
@ -876,7 +876,7 @@ Validating --> OL.t = Times (OL.W, HL2.y) : [132 x 512], [512 x 1 x *4] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *4], [132 x 1] -> [132 x 1 x *4]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *4], [132 x 1] -> [132 x 1 x *4]
|
||||
|
|
@ -901,7 +901,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 29 nodes to process in pass 1.
|
||||
|
|
@ -931,7 +931,7 @@ Validating --> OL.t = Times (OL.W, HL3.y) : [132 x 512], [512 x 1 x *4] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *4], [132 x 1] -> [132 x 1 x *4]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *4], [132 x 1] -> [132 x 1 x *4]
|
||||
|
|
@ -972,7 +972,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 29 nodes to process in pass 1.
|
||||
|
|
@ -1002,7 +1002,7 @@ Validating --> OL.t = Times (OL.W, HL3.y) : [132 x 512], [512 x 1 x *6] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *6], [132 x 1] -> [132 x 1 x *6]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *6], [132 x 1 x *6] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *6], [132 x 1 x *6] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *6], [132 x 1 x *6] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *6], [132 x 1] -> [132 x 1 x *6]
|
||||
|
|
@ -1024,7 +1024,7 @@ Post-processing network complete.
|
|||
08/16/2016 09:57:18: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 09:57:18: Evaluation criterion node(s):
|
||||
08/16/2016 09:57:18: err = ErrorPrediction
|
||||
08/16/2016 09:57:18: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -550,7 +550,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 1]) -> [512, MBSize 1]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 1]) -> [132, MBSize 1]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 1], OL.b[132, 1]) -> [132, MBSize 1]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
|
||||
Validating for node Err. 1 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -568,7 +568,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 1]) -> [512, MBSize 1]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 1]) -> [132, MBSize 1]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 1], OL.b[132, 1]) -> [132, MBSize 1]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -586,7 +586,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 1]) -> [512, MBSize 1]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 1]) -> [132, MBSize 1]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 1], OL.b[132, 1]) -> [132, MBSize 1]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
|
||||
7 out of 15 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -608,7 +608,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 1]) -> [512, MBSize 1]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 1]) -> [132, MBSize 1]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 1], OL.b[132, 1]) -> [132, MBSize 1]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -626,7 +626,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 1]) -> [512, MBSize 1]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 1]) -> [132, MBSize 1]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 1], OL.b[132, 1]) -> [132, MBSize 1]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 1], OL.z[132, MBSize 1]) -> [1, 1]
|
||||
|
||||
7 out of 15 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -947,7 +947,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err. 1 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -965,7 +965,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -983,7 +983,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
7 out of 15 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -1005,7 +1005,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -1023,7 +1023,7 @@ Validating --> HL1.y = Sigmoid(HL1.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL1.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
7 out of 15 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -1300,7 +1300,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -1323,7 +1323,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -1350,7 +1350,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -1373,7 +1373,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -1684,7 +1684,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err. 1 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -1707,7 +1707,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -1730,7 +1730,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -1757,7 +1757,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -1780,7 +1780,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -2161,7 +2161,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err. 1 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -2184,7 +2184,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -2207,7 +2207,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -2234,7 +2234,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -2257,7 +2257,7 @@ Validating --> HL2.y = Sigmoid(HL2.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL2.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
9 out of 20 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -2589,7 +2589,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -2617,7 +2617,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
11 out of 25 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -2649,7 +2649,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -2677,7 +2677,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
11 out of 25 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -3048,7 +3048,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err. 1 nodes to process in pass 2.
|
||||
|
||||
|
|
@ -3076,7 +3076,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -3104,7 +3104,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
11 out of 25 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
@ -3136,7 +3136,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
Validating for node Err, final verification.
|
||||
|
||||
|
|
@ -3164,7 +3164,7 @@ Validating --> HL3.y = Sigmoid(HL3.z[512, MBSize 0]) -> [512, MBSize 0]
|
|||
Validating --> OL.t = Times(OL.W[132, 512], HL3.y[512, MBSize 0]) -> [132, MBSize 0]
|
||||
Validating --> OL.b = LearnableParameter -> [132, 1]
|
||||
Validating --> OL.z = Plus(OL.t[132, MBSize 0], OL.b[132, 1]) -> [132, MBSize 0]
|
||||
Validating --> Err = ErrorPrediction(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
Validating --> Err = ClassificationError(labels[132, MBSize 0], OL.z[132, MBSize 0]) -> [1, 1]
|
||||
|
||||
11 out of 25 nodes do not share the minibatch layout with the input data.
|
||||
|
||||
|
|
|
|||
|
|
@ -392,7 +392,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 19 nodes to process in pass 1.
|
||||
|
|
@ -412,7 +412,7 @@ Validating --> OL.t = Times (OL.W, HL1.y) : [132 x 512], [512 x 1 x *] -> [132 x
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *], [132 x 1 x *] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
|
|
@ -434,7 +434,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:02:03: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:02:03: Evaluation criterion node(s):
|
||||
08/16/2016 03:02:03: err = ErrorPrediction
|
||||
08/16/2016 03:02:03: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -557,7 +557,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 19 nodes to process in pass 1.
|
||||
|
|
@ -577,7 +577,7 @@ Validating --> OL.t = Times (OL.W, HL1.y) : [132 x 512], [512 x 1 x *1] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *1], [132 x 1] -> [132 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *1], [132 x 1] -> [132 x 1 x *1]
|
||||
|
|
@ -602,7 +602,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 24 nodes to process in pass 1.
|
||||
|
|
@ -627,7 +627,7 @@ Validating --> OL.t = Times (OL.W, HL2.y) : [132 x 512], [512 x 1 x *1] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *1], [132 x 1] -> [132 x 1 x *1]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *1], [132 x 1 x *1] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *1], [132 x 1] -> [132 x 1 x *1]
|
||||
|
|
@ -668,7 +668,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 24 nodes to process in pass 1.
|
||||
|
|
@ -693,7 +693,7 @@ Validating --> OL.t = Times (OL.W, HL2.y) : [132 x 512], [512 x 1 x *3] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *3], [132 x 1] -> [132 x 1 x *3]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *3], [132 x 1 x *3] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *3], [132 x 1 x *3] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *3], [132 x 1 x *3] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *3], [132 x 1] -> [132 x 1 x *3]
|
||||
|
|
@ -715,7 +715,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:02:05: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:02:05: Evaluation criterion node(s):
|
||||
08/16/2016 03:02:05: err = ErrorPrediction
|
||||
08/16/2016 03:02:05: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -848,7 +848,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 24 nodes to process in pass 1.
|
||||
|
|
@ -873,7 +873,7 @@ Validating --> OL.t = Times (OL.W, HL2.y) : [132 x 512], [512 x 1 x *4] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *4], [132 x 1] -> [132 x 1 x *4]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *4], [132 x 1] -> [132 x 1 x *4]
|
||||
|
|
@ -898,7 +898,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 29 nodes to process in pass 1.
|
||||
|
|
@ -928,7 +928,7 @@ Validating --> OL.t = Times (OL.W, HL3.y) : [132 x 512], [512 x 1 x *4] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *4], [132 x 1] -> [132 x 1 x *4]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *4], [132 x 1 x *4] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *4], [132 x 1] -> [132 x 1 x *4]
|
||||
|
|
@ -969,7 +969,7 @@ Post-processing network...
|
|||
|
||||
3 roots:
|
||||
ce = CrossEntropyWithSoftmax()
|
||||
err = ErrorPrediction()
|
||||
err = ClassificationError()
|
||||
scaledLogLikelihood = Minus()
|
||||
|
||||
Validating network. 29 nodes to process in pass 1.
|
||||
|
|
@ -999,7 +999,7 @@ Validating --> OL.t = Times (OL.W, HL3.y) : [132 x 512], [512 x 1 x *6] -> [132
|
|||
Validating --> OL.b = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> OL.z = Plus (OL.t, OL.b) : [132 x 1 x *6], [132 x 1] -> [132 x 1 x *6]
|
||||
Validating --> ce = CrossEntropyWithSoftmax (labels, OL.z) : [132 x *6], [132 x 1 x *6] -> [1]
|
||||
Validating --> err = ErrorPrediction (labels, OL.z) : [132 x *6], [132 x 1 x *6] -> [1]
|
||||
Validating --> err = ClassificationError (labels, OL.z) : [132 x *6], [132 x 1 x *6] -> [1]
|
||||
Validating --> globalPrior = LearnableParameter() : -> [132 x 1]
|
||||
Validating --> logPrior = Log (globalPrior) : [132 x 1] -> [132 x 1]
|
||||
Validating --> scaledLogLikelihood = Minus (OL.z, logPrior) : [132 x 1 x *6], [132 x 1] -> [132 x 1 x *6]
|
||||
|
|
@ -1021,7 +1021,7 @@ Post-processing network complete.
|
|||
08/16/2016 03:02:08: ce = CrossEntropyWithSoftmax
|
||||
|
||||
08/16/2016 03:02:08: Evaluation criterion node(s):
|
||||
08/16/2016 03:02:08: err = ErrorPrediction
|
||||
08/16/2016 03:02:08: err = ClassificationError
|
||||
|
||||
|
||||
Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -105,7 +105,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -194,7 +194,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -289,7 +289,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -402,7 +402,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 6 roots:
|
||||
MPI Rank 0: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: err = ErrorPrediction()
|
||||
MPI Rank 0: err = ClassificationError()
|
||||
MPI Rank 0: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 0: featNorm.mean = Mean()
|
||||
MPI Rank 0: logPrior._ = Mean()
|
||||
|
|
@ -443,7 +443,7 @@ MPI Rank 0: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 0: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 0: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -465,7 +465,7 @@ MPI Rank 0: 08/16/2016 09:57:23: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 09:57:23: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:57:23: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 09:57:23: err = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 09:57:23: err = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -658,7 +658,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -747,7 +747,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -842,7 +842,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -955,7 +955,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 6 roots:
|
||||
MPI Rank 1: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: err = ErrorPrediction()
|
||||
MPI Rank 1: err = ClassificationError()
|
||||
MPI Rank 1: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 1: featNorm.mean = Mean()
|
||||
MPI Rank 1: logPrior._ = Mean()
|
||||
|
|
@ -996,7 +996,7 @@ MPI Rank 1: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 1: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 1: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -1018,7 +1018,7 @@ MPI Rank 1: 08/16/2016 09:57:23: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 09:57:23: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:57:23: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 09:57:23: err = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 09:57:23: err = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1265,7 +1265,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1354,7 +1354,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1449,7 +1449,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1544,7 +1544,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 6 roots:
|
||||
MPI Rank 0: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: err = ErrorPrediction()
|
||||
MPI Rank 0: err = ClassificationError()
|
||||
MPI Rank 0: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 0: featNorm.mean = Mean()
|
||||
MPI Rank 0: logPrior._ = Mean()
|
||||
|
|
@ -1585,7 +1585,7 @@ MPI Rank 0: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 0: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 0: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -1607,7 +1607,7 @@ MPI Rank 0: 08/16/2016 09:58:04: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 09:58:04: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:58:04: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 09:58:04: err = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 09:58:04: err = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1745,7 +1745,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -1834,7 +1834,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -1929,7 +1929,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -2024,7 +2024,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 6 roots:
|
||||
MPI Rank 1: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: err = ErrorPrediction()
|
||||
MPI Rank 1: err = ClassificationError()
|
||||
MPI Rank 1: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 1: featNorm.mean = Mean()
|
||||
MPI Rank 1: logPrior._ = Mean()
|
||||
|
|
@ -2065,7 +2065,7 @@ MPI Rank 1: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 1: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 1: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -2087,7 +2087,7 @@ MPI Rank 1: 08/16/2016 09:58:04: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 09:58:04: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:04: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 09:58:04: err = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 09:58:04: err = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -105,7 +105,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -194,7 +194,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -289,7 +289,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -402,7 +402,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 6 roots:
|
||||
MPI Rank 0: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: err = ErrorPrediction()
|
||||
MPI Rank 0: err = ClassificationError()
|
||||
MPI Rank 0: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 0: featNorm.mean = Mean()
|
||||
MPI Rank 0: logPrior._ = Mean()
|
||||
|
|
@ -443,7 +443,7 @@ MPI Rank 0: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 0: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 0: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -465,7 +465,7 @@ MPI Rank 0: 08/16/2016 09:58:22: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 09:58:22: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:58:22: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 09:58:22: err = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 09:58:22: err = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -674,7 +674,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -763,7 +763,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -858,7 +858,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -971,7 +971,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 6 roots:
|
||||
MPI Rank 1: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: err = ErrorPrediction()
|
||||
MPI Rank 1: err = ClassificationError()
|
||||
MPI Rank 1: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 1: featNorm.mean = Mean()
|
||||
MPI Rank 1: logPrior._ = Mean()
|
||||
|
|
@ -1012,7 +1012,7 @@ MPI Rank 1: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 1: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 1: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -1034,7 +1034,7 @@ MPI Rank 1: 08/16/2016 09:58:23: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 09:58:23: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:23: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 09:58:23: err = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 09:58:23: err = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1297,7 +1297,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1386,7 +1386,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1481,7 +1481,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1576,7 +1576,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 6 roots:
|
||||
MPI Rank 0: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: err = ErrorPrediction()
|
||||
MPI Rank 0: err = ClassificationError()
|
||||
MPI Rank 0: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 0: featNorm.mean = Mean()
|
||||
MPI Rank 0: logPrior._ = Mean()
|
||||
|
|
@ -1617,7 +1617,7 @@ MPI Rank 0: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 0: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 0: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -1639,7 +1639,7 @@ MPI Rank 0: 08/16/2016 09:58:31: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 09:58:31: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:58:31: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 09:58:31: err = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 09:58:31: err = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1781,7 +1781,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -1870,7 +1870,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -1965,7 +1965,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -2060,7 +2060,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 6 roots:
|
||||
MPI Rank 1: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: err = ErrorPrediction()
|
||||
MPI Rank 1: err = ClassificationError()
|
||||
MPI Rank 1: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 1: featNorm.mean = Mean()
|
||||
MPI Rank 1: logPrior._ = Mean()
|
||||
|
|
@ -2101,7 +2101,7 @@ MPI Rank 1: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 1: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 1: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -2123,7 +2123,7 @@ MPI Rank 1: 08/16/2016 09:58:31: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 09:58:31: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:31: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 09:58:31: err = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 09:58:31: err = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -101,7 +101,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -190,7 +190,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -285,7 +285,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -398,7 +398,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 6 roots:
|
||||
MPI Rank 0: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: err = ErrorPrediction()
|
||||
MPI Rank 0: err = ClassificationError()
|
||||
MPI Rank 0: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 0: featNorm.mean = Mean()
|
||||
MPI Rank 0: logPrior._ = Mean()
|
||||
|
|
@ -439,7 +439,7 @@ MPI Rank 0: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 0: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 0: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -461,7 +461,7 @@ MPI Rank 0: 08/16/2016 03:02:17: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:02:17: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:17: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:02:17: err = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:02:17: err = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -652,7 +652,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -741,7 +741,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -836,7 +836,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -949,7 +949,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 6 roots:
|
||||
MPI Rank 1: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: err = ErrorPrediction()
|
||||
MPI Rank 1: err = ClassificationError()
|
||||
MPI Rank 1: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 1: featNorm.mean = Mean()
|
||||
MPI Rank 1: logPrior._ = Mean()
|
||||
|
|
@ -990,7 +990,7 @@ MPI Rank 1: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 1: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 1: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -1012,7 +1012,7 @@ MPI Rank 1: 08/16/2016 03:02:18: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:02:18: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:18: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:02:18: err = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:02:18: err = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1255,7 +1255,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1344,7 +1344,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1439,7 +1439,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1534,7 +1534,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 6 roots:
|
||||
MPI Rank 0: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: err = ErrorPrediction()
|
||||
MPI Rank 0: err = ClassificationError()
|
||||
MPI Rank 0: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 0: featNorm.mean = Mean()
|
||||
MPI Rank 0: logPrior._ = Mean()
|
||||
|
|
@ -1575,7 +1575,7 @@ MPI Rank 0: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 0: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 0: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -1597,7 +1597,7 @@ MPI Rank 0: 08/16/2016 03:03:00: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:03:00: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:03:00: err = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:03:00: err = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1733,7 +1733,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -1822,7 +1822,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -1917,7 +1917,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -2012,7 +2012,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 6 roots:
|
||||
MPI Rank 1: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: err = ErrorPrediction()
|
||||
MPI Rank 1: err = ClassificationError()
|
||||
MPI Rank 1: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 1: featNorm.mean = Mean()
|
||||
MPI Rank 1: logPrior._ = Mean()
|
||||
|
|
@ -2053,7 +2053,7 @@ MPI Rank 1: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 1: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 1: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -2075,7 +2075,7 @@ MPI Rank 1: 08/16/2016 03:03:01: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:03:01: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:03:01: err = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:03:01: err = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -101,7 +101,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -190,7 +190,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -285,7 +285,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -398,7 +398,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 6 roots:
|
||||
MPI Rank 0: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: err = ErrorPrediction()
|
||||
MPI Rank 0: err = ClassificationError()
|
||||
MPI Rank 0: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 0: featNorm.mean = Mean()
|
||||
MPI Rank 0: logPrior._ = Mean()
|
||||
|
|
@ -439,7 +439,7 @@ MPI Rank 0: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 0: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 0: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -461,7 +461,7 @@ MPI Rank 0: 08/16/2016 03:03:21: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:03:21: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:21: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:03:21: err = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:03:21: err = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -668,7 +668,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -757,7 +757,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -852,7 +852,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -965,7 +965,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 6 roots:
|
||||
MPI Rank 1: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: err = ErrorPrediction()
|
||||
MPI Rank 1: err = ClassificationError()
|
||||
MPI Rank 1: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 1: featNorm.mean = Mean()
|
||||
MPI Rank 1: logPrior._ = Mean()
|
||||
|
|
@ -1006,7 +1006,7 @@ MPI Rank 1: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 1: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 1: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -1028,7 +1028,7 @@ MPI Rank 1: 08/16/2016 03:03:21: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:03:21: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:21: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:03:21: err = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:03:21: err = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1287,7 +1287,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1376,7 +1376,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1471,7 +1471,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 0: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 0: evalCriterion=ErrorPrediction
|
||||
MPI Rank 0: evalCriterion=ClassificationError
|
||||
MPI Rank 0: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
MPI Rank 0: initValueScale=1.0
|
||||
|
|
@ -1566,7 +1566,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 6 roots:
|
||||
MPI Rank 0: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: err = ErrorPrediction()
|
||||
MPI Rank 0: err = ClassificationError()
|
||||
MPI Rank 0: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 0: featNorm.mean = Mean()
|
||||
MPI Rank 0: logPrior._ = Mean()
|
||||
|
|
@ -1607,7 +1607,7 @@ MPI Rank 0: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 0: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 0: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -1629,7 +1629,7 @@ MPI Rank 0: 08/16/2016 03:03:32: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:03:32: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:32: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:03:32: err = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:03:32: err = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1769,7 +1769,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -1858,7 +1858,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -1953,7 +1953,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: BrainScriptNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:512:132
|
||||
MPI Rank 1: trainingCriterion=CrossEntropyWithSoftmax
|
||||
MPI Rank 1: evalCriterion=ErrorPrediction
|
||||
MPI Rank 1: evalCriterion=ClassificationError
|
||||
MPI Rank 1: layerTypes[i:1..Length(layerSizes)-2]=Sigmoid
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
MPI Rank 1: initValueScale=1.0
|
||||
|
|
@ -2048,7 +2048,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 6 roots:
|
||||
MPI Rank 1: ce = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: err = ErrorPrediction()
|
||||
MPI Rank 1: err = ClassificationError()
|
||||
MPI Rank 1: featNorm.invStdDev = InvStdDev()
|
||||
MPI Rank 1: featNorm.mean = Mean()
|
||||
MPI Rank 1: logPrior._ = Mean()
|
||||
|
|
@ -2089,7 +2089,7 @@ MPI Rank 1: Validating --> outLayer.z.PlusArgs[0] = Times (outLayer.W, outLayer.
|
|||
MPI Rank 1: Validating --> outLayer.B = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> outZ = Plus (outLayer.z.PlusArgs[0], outLayer.B) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> ce = CrossEntropyWithSoftmax (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ErrorPrediction (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> err = ClassificationError (labels, outZ) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> logPrior._ = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> logPrior = Log (logPrior._) : [132] -> [132]
|
||||
MPI Rank 1: Validating --> scaledLogLikelihood = Minus (outZ, logPrior) : [132 x 1 x *], [132] -> [132 x 1 x *]
|
||||
|
|
@ -2111,7 +2111,7 @@ MPI Rank 1: 08/16/2016 03:03:33: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:03:33: ce = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:33: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:03:33: err = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:03:33: err = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
|
|||
|
|
@ -133,7 +133,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -159,7 +159,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -239,7 +239,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -265,7 +265,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -351,7 +351,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -377,7 +377,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -473,7 +473,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -502,7 +502,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -525,7 +525,7 @@ MPI Rank 0: 08/16/2016 10:07:39: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 10:07:39: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:07:39: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 10:07:39: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 10:07:39: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -578,63 +578,63 @@ MPI Rank 0: 08/16/2016 10:07:42: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:07:42: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.1852s; samplesPerSecond = 3455.3
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.3127s; samplesPerSecond = 2047.0
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.1247s; samplesPerSecond = 5130.8
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1002s; samplesPerSecond = 6389.6
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.1956s; samplesPerSecond = 3272.6
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.1096s; samplesPerSecond = 5836.9
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.1876s; samplesPerSecond = 3411.2
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.1501s; samplesPerSecond = 4264.4
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.2478s; samplesPerSecond = 2582.4
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.1527s; samplesPerSecond = 4191.2
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.1190s; samplesPerSecond = 5377.0
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.1197s; samplesPerSecond = 5346.2
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.1199s; samplesPerSecond = 5335.7
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.1129s; samplesPerSecond = 5667.9
|
||||
MPI Rank 0: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.2553s; samplesPerSecond = 2506.9
|
||||
MPI Rank 0: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.1417s; samplesPerSecond = 4517.3
|
||||
MPI Rank 0: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.2422s; samplesPerSecond = 2642.2
|
||||
MPI Rank 0: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.2591s; samplesPerSecond = 2470.2
|
||||
MPI Rank 0: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.2458s; samplesPerSecond = 2603.2
|
||||
MPI Rank 0: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.4506s; samplesPerSecond = 1420.3
|
||||
MPI Rank 0: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.2081s; samplesPerSecond = 3075.6
|
||||
MPI Rank 0: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.2185s; samplesPerSecond = 2929.6
|
||||
MPI Rank 0: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2135s; samplesPerSecond = 2998.0
|
||||
MPI Rank 0: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.3871s; samplesPerSecond = 1653.5
|
||||
MPI Rank 0: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.1749s; samplesPerSecond = 3658.8
|
||||
MPI Rank 0: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.0950s; samplesPerSecond = 6734.7
|
||||
MPI Rank 0: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.1075s; samplesPerSecond = 5956.2
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1185s; samplesPerSecond = 5401.5
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.1115s; samplesPerSecond = 5742.1
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0978s; samplesPerSecond = 6542.2
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.1950s; samplesPerSecond = 3281.9
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.1188s; samplesPerSecond = 5386.8
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.88422s
|
||||
MPI Rank 0: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.1852s; samplesPerSecond = 3455.3
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.3127s; samplesPerSecond = 2047.0
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.1247s; samplesPerSecond = 5130.8
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1002s; samplesPerSecond = 6389.6
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.1956s; samplesPerSecond = 3272.6
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.1096s; samplesPerSecond = 5836.9
|
||||
MPI Rank 0: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.1876s; samplesPerSecond = 3411.2
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.1501s; samplesPerSecond = 4264.4
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.2478s; samplesPerSecond = 2582.4
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.1527s; samplesPerSecond = 4191.2
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.1190s; samplesPerSecond = 5377.0
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.1197s; samplesPerSecond = 5346.2
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.1199s; samplesPerSecond = 5335.7
|
||||
MPI Rank 0: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.1129s; samplesPerSecond = 5667.9
|
||||
MPI Rank 0: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.2553s; samplesPerSecond = 2506.9
|
||||
MPI Rank 0: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.1417s; samplesPerSecond = 4517.3
|
||||
MPI Rank 0: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.2422s; samplesPerSecond = 2642.2
|
||||
MPI Rank 0: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.2591s; samplesPerSecond = 2470.2
|
||||
MPI Rank 0: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.2458s; samplesPerSecond = 2603.2
|
||||
MPI Rank 0: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.4506s; samplesPerSecond = 1420.3
|
||||
MPI Rank 0: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.2081s; samplesPerSecond = 3075.6
|
||||
MPI Rank 0: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.2185s; samplesPerSecond = 2929.6
|
||||
MPI Rank 0: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2135s; samplesPerSecond = 2998.0
|
||||
MPI Rank 0: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.3871s; samplesPerSecond = 1653.5
|
||||
MPI Rank 0: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.1749s; samplesPerSecond = 3658.8
|
||||
MPI Rank 0: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.0950s; samplesPerSecond = 6734.7
|
||||
MPI Rank 0: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.1075s; samplesPerSecond = 5956.2
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1185s; samplesPerSecond = 5401.5
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.1115s; samplesPerSecond = 5742.1
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0978s; samplesPerSecond = 6542.2
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.1950s; samplesPerSecond = 3281.9
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.1188s; samplesPerSecond = 5386.8
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.88422s
|
||||
MPI Rank 0: 08/16/2016 10:07:48: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_Parallel1BitQuantization@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:07:48: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:07:49: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17725069 * 2560; EvalErrorPrediction = 0.59921875 * 2560; time = 0.5690s; samplesPerSecond = 4499.5
|
||||
MPI Rank 0: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.13334089 * 2560; EvalErrorPrediction = 0.56835938 * 2560; time = 0.7981s; samplesPerSecond = 3207.4
|
||||
MPI Rank 0: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07967076 * 2560; EvalErrorPrediction = 0.56328125 * 2560; time = 0.4440s; samplesPerSecond = 5765.9
|
||||
MPI Rank 0: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.04241663 * 2560; EvalErrorPrediction = 0.56171875 * 2560; time = 0.7284s; samplesPerSecond = 3514.6
|
||||
MPI Rank 0: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04365452 * 2560; EvalErrorPrediction = 0.55546875 * 2560; time = 0.7579s; samplesPerSecond = 3377.6
|
||||
MPI Rank 0: 08/16/2016 10:07:52: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.98165284 * 2560; EvalErrorPrediction = 0.54687500 * 2560; time = 0.4371s; samplesPerSecond = 5856.7
|
||||
MPI Rank 0: 08/16/2016 10:07:52: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.97900531 * 2560; EvalErrorPrediction = 0.53359375 * 2560; time = 0.5896s; samplesPerSecond = 4342.2
|
||||
MPI Rank 0: 08/16/2016 10:07:53: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.95254162 * 2560; EvalErrorPrediction = 0.54179687 * 2560; time = 0.5246s; samplesPerSecond = 4879.6
|
||||
MPI Rank 0: 08/16/2016 10:07:53: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04869166 * 20480; EvalErrorPrediction = 0.55878906 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.86648s
|
||||
MPI Rank 0: 08/16/2016 10:07:49: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17725069 * 2560; EvalClassificationError = 0.59921875 * 2560; time = 0.5690s; samplesPerSecond = 4499.5
|
||||
MPI Rank 0: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.13334089 * 2560; EvalClassificationError = 0.56835938 * 2560; time = 0.7981s; samplesPerSecond = 3207.4
|
||||
MPI Rank 0: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07967076 * 2560; EvalClassificationError = 0.56328125 * 2560; time = 0.4440s; samplesPerSecond = 5765.9
|
||||
MPI Rank 0: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.04241663 * 2560; EvalClassificationError = 0.56171875 * 2560; time = 0.7284s; samplesPerSecond = 3514.6
|
||||
MPI Rank 0: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04365452 * 2560; EvalClassificationError = 0.55546875 * 2560; time = 0.7579s; samplesPerSecond = 3377.6
|
||||
MPI Rank 0: 08/16/2016 10:07:52: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.98165284 * 2560; EvalClassificationError = 0.54687500 * 2560; time = 0.4371s; samplesPerSecond = 5856.7
|
||||
MPI Rank 0: 08/16/2016 10:07:52: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.97900531 * 2560; EvalClassificationError = 0.53359375 * 2560; time = 0.5896s; samplesPerSecond = 4342.2
|
||||
MPI Rank 0: 08/16/2016 10:07:53: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.95254162 * 2560; EvalClassificationError = 0.54179687 * 2560; time = 0.5246s; samplesPerSecond = 4879.6
|
||||
MPI Rank 0: 08/16/2016 10:07:53: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04869166 * 20480; EvalClassificationError = 0.55878906 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.86648s
|
||||
MPI Rank 0: 08/16/2016 10:07:53: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_Parallel1BitQuantization@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:07:53: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:07:53: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:07:54: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97049696 * 10240; EvalErrorPrediction = 0.54658203 * 10240; time = 1.3527s; samplesPerSecond = 7570.1
|
||||
MPI Rank 0: 08/16/2016 10:07:56: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96995673 * 10240; EvalErrorPrediction = 0.54746094 * 10240; time = 1.6866s; samplesPerSecond = 6071.4
|
||||
MPI Rank 0: 08/16/2016 10:07:56: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.97022685 * 20480; EvalErrorPrediction = 0.54702148 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.0565s
|
||||
MPI Rank 0: 08/16/2016 10:07:54: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97049696 * 10240; EvalClassificationError = 0.54658203 * 10240; time = 1.3527s; samplesPerSecond = 7570.1
|
||||
MPI Rank 0: 08/16/2016 10:07:56: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96995673 * 10240; EvalClassificationError = 0.54746094 * 10240; time = 1.6866s; samplesPerSecond = 6071.4
|
||||
MPI Rank 0: 08/16/2016 10:07:56: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.97022685 * 20480; EvalClassificationError = 0.54702148 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.0565s
|
||||
MPI Rank 0: 08/16/2016 10:07:56: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_Parallel1BitQuantization@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 10:07:56: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -687,7 +687,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -713,7 +713,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -793,7 +793,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -819,7 +819,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -905,7 +905,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -931,7 +931,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1027,7 +1027,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1056,7 +1056,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1079,7 +1079,7 @@ MPI Rank 1: 08/16/2016 10:07:39: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 10:07:39: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:07:39: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 10:07:39: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 10:07:39: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1132,61 +1132,61 @@ MPI Rank 1: 08/16/2016 10:07:42: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.0820s; samplesPerSecond = 7802.3
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.0526s; samplesPerSecond = 12174.0
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0521s; samplesPerSecond = 12274.6
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0526s; samplesPerSecond = 12171.0
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.0526s; samplesPerSecond = 12176.3
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.0521s; samplesPerSecond = 12275.8
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.0522s; samplesPerSecond = 12260.5
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.0522s; samplesPerSecond = 12257.2
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.0525s; samplesPerSecond = 12197.7
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.0524s; samplesPerSecond = 12215.8
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.0524s; samplesPerSecond = 12223.8
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.0524s; samplesPerSecond = 12214.2
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0522s; samplesPerSecond = 12249.0
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.0524s; samplesPerSecond = 12221.7
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.0545s; samplesPerSecond = 11744.6
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.0521s; samplesPerSecond = 12273.5
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.0527s; samplesPerSecond = 12138.2
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.0523s; samplesPerSecond = 12241.3
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.0520s; samplesPerSecond = 12312.2
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.0521s; samplesPerSecond = 12288.1
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0530s; samplesPerSecond = 12064.8
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.0524s; samplesPerSecond = 12222.1
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0526s; samplesPerSecond = 12174.7
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0521s; samplesPerSecond = 12280.1
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0526s; samplesPerSecond = 12168.7
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.0522s; samplesPerSecond = 12250.2
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.0523s; samplesPerSecond = 12239.2
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0526s; samplesPerSecond = 12159.4
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.0531s; samplesPerSecond = 12051.1
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0526s; samplesPerSecond = 12162.9
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.0538s; samplesPerSecond = 11906.1
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.0522s; samplesPerSecond = 12256.8
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.71524s
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.0820s; samplesPerSecond = 7802.3
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.0526s; samplesPerSecond = 12174.0
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0521s; samplesPerSecond = 12274.6
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0526s; samplesPerSecond = 12171.0
|
||||
MPI Rank 1: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.0526s; samplesPerSecond = 12176.3
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.0521s; samplesPerSecond = 12275.8
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.0522s; samplesPerSecond = 12260.5
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.0522s; samplesPerSecond = 12257.2
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.0525s; samplesPerSecond = 12197.7
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.0524s; samplesPerSecond = 12215.8
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.0524s; samplesPerSecond = 12223.8
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.0524s; samplesPerSecond = 12214.2
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0522s; samplesPerSecond = 12249.0
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.0524s; samplesPerSecond = 12221.7
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.0545s; samplesPerSecond = 11744.6
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.0521s; samplesPerSecond = 12273.5
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.0527s; samplesPerSecond = 12138.2
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.0523s; samplesPerSecond = 12241.3
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.0520s; samplesPerSecond = 12312.2
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.0521s; samplesPerSecond = 12288.1
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0530s; samplesPerSecond = 12064.8
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.0524s; samplesPerSecond = 12222.1
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0526s; samplesPerSecond = 12174.7
|
||||
MPI Rank 1: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0521s; samplesPerSecond = 12280.1
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0526s; samplesPerSecond = 12168.7
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.0522s; samplesPerSecond = 12250.2
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.0523s; samplesPerSecond = 12239.2
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0526s; samplesPerSecond = 12159.4
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.0531s; samplesPerSecond = 12051.1
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0526s; samplesPerSecond = 12162.9
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.0538s; samplesPerSecond = 11906.1
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.0522s; samplesPerSecond = 12256.8
|
||||
MPI Rank 1: 08/16/2016 10:07:44: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.71524s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:07:48: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:07:48: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:07:49: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17725069 * 2560; EvalErrorPrediction = 0.59921875 * 2560; time = 0.5643s; samplesPerSecond = 4537.0
|
||||
MPI Rank 1: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.13334089 * 2560; EvalErrorPrediction = 0.56835938 * 2560; time = 0.7961s; samplesPerSecond = 3215.7
|
||||
MPI Rank 1: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07967076 * 2560; EvalErrorPrediction = 0.56328125 * 2560; time = 0.4504s; samplesPerSecond = 5684.3
|
||||
MPI Rank 1: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.04241663 * 2560; EvalErrorPrediction = 0.56171875 * 2560; time = 0.7289s; samplesPerSecond = 3512.2
|
||||
MPI Rank 1: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04365452 * 2560; EvalErrorPrediction = 0.55546875 * 2560; time = 0.7567s; samplesPerSecond = 3383.2
|
||||
MPI Rank 1: 08/16/2016 10:07:52: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.98165284 * 2560; EvalErrorPrediction = 0.54687500 * 2560; time = 0.4379s; samplesPerSecond = 5846.7
|
||||
MPI Rank 1: 08/16/2016 10:07:52: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.97900531 * 2560; EvalErrorPrediction = 0.53359375 * 2560; time = 0.5895s; samplesPerSecond = 4342.3
|
||||
MPI Rank 1: 08/16/2016 10:07:53: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.95254162 * 2560; EvalErrorPrediction = 0.54179687 * 2560; time = 0.5098s; samplesPerSecond = 5021.3
|
||||
MPI Rank 1: 08/16/2016 10:07:53: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04869166 * 20480; EvalErrorPrediction = 0.55878906 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.86648s
|
||||
MPI Rank 1: 08/16/2016 10:07:49: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17725069 * 2560; EvalClassificationError = 0.59921875 * 2560; time = 0.5643s; samplesPerSecond = 4537.0
|
||||
MPI Rank 1: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.13334089 * 2560; EvalClassificationError = 0.56835938 * 2560; time = 0.7961s; samplesPerSecond = 3215.7
|
||||
MPI Rank 1: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07967076 * 2560; EvalClassificationError = 0.56328125 * 2560; time = 0.4504s; samplesPerSecond = 5684.3
|
||||
MPI Rank 1: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.04241663 * 2560; EvalClassificationError = 0.56171875 * 2560; time = 0.7289s; samplesPerSecond = 3512.2
|
||||
MPI Rank 1: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04365452 * 2560; EvalClassificationError = 0.55546875 * 2560; time = 0.7567s; samplesPerSecond = 3383.2
|
||||
MPI Rank 1: 08/16/2016 10:07:52: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.98165284 * 2560; EvalClassificationError = 0.54687500 * 2560; time = 0.4379s; samplesPerSecond = 5846.7
|
||||
MPI Rank 1: 08/16/2016 10:07:52: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.97900531 * 2560; EvalClassificationError = 0.53359375 * 2560; time = 0.5895s; samplesPerSecond = 4342.3
|
||||
MPI Rank 1: 08/16/2016 10:07:53: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.95254162 * 2560; EvalClassificationError = 0.54179687 * 2560; time = 0.5098s; samplesPerSecond = 5021.3
|
||||
MPI Rank 1: 08/16/2016 10:07:53: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04869166 * 20480; EvalClassificationError = 0.55878906 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.86648s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:07:53: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:07:53: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:07:54: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97049696 * 10240; EvalErrorPrediction = 0.54658203 * 10240; time = 1.3515s; samplesPerSecond = 7576.5
|
||||
MPI Rank 1: 08/16/2016 10:07:56: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96995673 * 10240; EvalErrorPrediction = 0.54746094 * 10240; time = 1.6863s; samplesPerSecond = 6072.6
|
||||
MPI Rank 1: 08/16/2016 10:07:56: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.97022685 * 20480; EvalErrorPrediction = 0.54702148 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.05591s
|
||||
MPI Rank 1: 08/16/2016 10:07:54: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97049696 * 10240; EvalClassificationError = 0.54658203 * 10240; time = 1.3515s; samplesPerSecond = 7576.5
|
||||
MPI Rank 1: 08/16/2016 10:07:56: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96995673 * 10240; EvalClassificationError = 0.54746094 * 10240; time = 1.6863s; samplesPerSecond = 6072.6
|
||||
MPI Rank 1: 08/16/2016 10:07:56: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.97022685 * 20480; EvalClassificationError = 0.54702148 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.05591s
|
||||
MPI Rank 1: 08/16/2016 10:07:56: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:07:56: Action "train" complete.
|
||||
|
|
@ -1238,7 +1238,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1264,7 +1264,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1344,7 +1344,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1370,7 +1370,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1456,7 +1456,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1482,7 +1482,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1578,7 +1578,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1607,7 +1607,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1630,7 +1630,7 @@ MPI Rank 2: 08/16/2016 10:07:40: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 10:07:40: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:07:40: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 10:07:40: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 10:07:40: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1683,61 +1683,61 @@ MPI Rank 2: 08/16/2016 10:07:42: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:07:42: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.2713s; samplesPerSecond = 2359.3
|
||||
MPI Rank 2: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.2312s; samplesPerSecond = 2768.6
|
||||
MPI Rank 2: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.3000s; samplesPerSecond = 2133.2
|
||||
MPI Rank 2: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1609s; samplesPerSecond = 3978.8
|
||||
MPI Rank 2: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.1049s; samplesPerSecond = 6103.0
|
||||
MPI Rank 2: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.1162s; samplesPerSecond = 5508.7
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.2419s; samplesPerSecond = 2645.3
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.2571s; samplesPerSecond = 2489.6
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.1202s; samplesPerSecond = 5322.5
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.1196s; samplesPerSecond = 5352.9
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.1208s; samplesPerSecond = 5300.0
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.2270s; samplesPerSecond = 2819.0
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.1069s; samplesPerSecond = 5985.6
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.3318s; samplesPerSecond = 1928.9
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.0973s; samplesPerSecond = 6574.9
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.1151s; samplesPerSecond = 5559.1
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.1552s; samplesPerSecond = 4122.6
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.1163s; samplesPerSecond = 5505.0
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.0965s; samplesPerSecond = 6633.3
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.1497s; samplesPerSecond = 4274.9
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1479s; samplesPerSecond = 4326.2
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.2516s; samplesPerSecond = 2544.1
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2075s; samplesPerSecond = 3084.6
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.1087s; samplesPerSecond = 5886.5
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.1703s; samplesPerSecond = 3758.8
|
||||
MPI Rank 2: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.2147s; samplesPerSecond = 2981.0
|
||||
MPI Rank 2: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.2963s; samplesPerSecond = 2160.1
|
||||
MPI Rank 2: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.2348s; samplesPerSecond = 2725.6
|
||||
MPI Rank 2: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.1425s; samplesPerSecond = 4491.5
|
||||
MPI Rank 2: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.1256s; samplesPerSecond = 5096.0
|
||||
MPI Rank 2: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.2974s; samplesPerSecond = 2151.8
|
||||
MPI Rank 2: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.1329s; samplesPerSecond = 4814.1
|
||||
MPI Rank 2: 08/16/2016 10:07:48: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.77603s
|
||||
MPI Rank 2: 08/16/2016 10:07:42: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.2713s; samplesPerSecond = 2359.3
|
||||
MPI Rank 2: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.2312s; samplesPerSecond = 2768.6
|
||||
MPI Rank 2: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.3000s; samplesPerSecond = 2133.2
|
||||
MPI Rank 2: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1609s; samplesPerSecond = 3978.8
|
||||
MPI Rank 2: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.1049s; samplesPerSecond = 6103.0
|
||||
MPI Rank 2: 08/16/2016 10:07:43: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.1162s; samplesPerSecond = 5508.7
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.2419s; samplesPerSecond = 2645.3
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.2571s; samplesPerSecond = 2489.6
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.1202s; samplesPerSecond = 5322.5
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.1196s; samplesPerSecond = 5352.9
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.1208s; samplesPerSecond = 5300.0
|
||||
MPI Rank 2: 08/16/2016 10:07:44: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.2270s; samplesPerSecond = 2819.0
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.1069s; samplesPerSecond = 5985.6
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.3318s; samplesPerSecond = 1928.9
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.0973s; samplesPerSecond = 6574.9
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.1151s; samplesPerSecond = 5559.1
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.1552s; samplesPerSecond = 4122.6
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.1163s; samplesPerSecond = 5505.0
|
||||
MPI Rank 2: 08/16/2016 10:07:45: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.0965s; samplesPerSecond = 6633.3
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.1497s; samplesPerSecond = 4274.9
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1479s; samplesPerSecond = 4326.2
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.2516s; samplesPerSecond = 2544.1
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2075s; samplesPerSecond = 3084.6
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.1087s; samplesPerSecond = 5886.5
|
||||
MPI Rank 2: 08/16/2016 10:07:46: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.1703s; samplesPerSecond = 3758.8
|
||||
MPI Rank 2: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.2147s; samplesPerSecond = 2981.0
|
||||
MPI Rank 2: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.2963s; samplesPerSecond = 2160.1
|
||||
MPI Rank 2: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.2348s; samplesPerSecond = 2725.6
|
||||
MPI Rank 2: 08/16/2016 10:07:47: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.1425s; samplesPerSecond = 4491.5
|
||||
MPI Rank 2: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.1256s; samplesPerSecond = 5096.0
|
||||
MPI Rank 2: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.2974s; samplesPerSecond = 2151.8
|
||||
MPI Rank 2: 08/16/2016 10:07:48: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.1329s; samplesPerSecond = 4814.1
|
||||
MPI Rank 2: 08/16/2016 10:07:48: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.77603s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:07:48: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:07:48: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 10:07:49: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17725069 * 2560; EvalErrorPrediction = 0.59921875 * 2560; time = 0.5710s; samplesPerSecond = 4483.5
|
||||
MPI Rank 2: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.13334089 * 2560; EvalErrorPrediction = 0.56835938 * 2560; time = 0.7894s; samplesPerSecond = 3242.9
|
||||
MPI Rank 2: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07967076 * 2560; EvalErrorPrediction = 0.56328125 * 2560; time = 0.4520s; samplesPerSecond = 5664.3
|
||||
MPI Rank 2: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.04241663 * 2560; EvalErrorPrediction = 0.56171875 * 2560; time = 0.7336s; samplesPerSecond = 3489.6
|
||||
MPI Rank 2: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04365452 * 2560; EvalErrorPrediction = 0.55546875 * 2560; time = 0.7518s; samplesPerSecond = 3405.1
|
||||
MPI Rank 2: 08/16/2016 10:07:52: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.98165284 * 2560; EvalErrorPrediction = 0.54687500 * 2560; time = 0.4396s; samplesPerSecond = 5823.5
|
||||
MPI Rank 2: 08/16/2016 10:07:53: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.97900531 * 2560; EvalErrorPrediction = 0.53359375 * 2560; time = 0.5940s; samplesPerSecond = 4309.7
|
||||
MPI Rank 2: 08/16/2016 10:07:53: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.95254162 * 2560; EvalErrorPrediction = 0.54179687 * 2560; time = 0.5104s; samplesPerSecond = 5016.0
|
||||
MPI Rank 2: 08/16/2016 10:07:53: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04869166 * 20480; EvalErrorPrediction = 0.55878906 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.87459s
|
||||
MPI Rank 2: 08/16/2016 10:07:49: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17725069 * 2560; EvalClassificationError = 0.59921875 * 2560; time = 0.5710s; samplesPerSecond = 4483.5
|
||||
MPI Rank 2: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.13334089 * 2560; EvalClassificationError = 0.56835938 * 2560; time = 0.7894s; samplesPerSecond = 3242.9
|
||||
MPI Rank 2: 08/16/2016 10:07:50: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07967076 * 2560; EvalClassificationError = 0.56328125 * 2560; time = 0.4520s; samplesPerSecond = 5664.3
|
||||
MPI Rank 2: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.04241663 * 2560; EvalClassificationError = 0.56171875 * 2560; time = 0.7336s; samplesPerSecond = 3489.6
|
||||
MPI Rank 2: 08/16/2016 10:07:51: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04365452 * 2560; EvalClassificationError = 0.55546875 * 2560; time = 0.7518s; samplesPerSecond = 3405.1
|
||||
MPI Rank 2: 08/16/2016 10:07:52: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.98165284 * 2560; EvalClassificationError = 0.54687500 * 2560; time = 0.4396s; samplesPerSecond = 5823.5
|
||||
MPI Rank 2: 08/16/2016 10:07:53: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.97900531 * 2560; EvalClassificationError = 0.53359375 * 2560; time = 0.5940s; samplesPerSecond = 4309.7
|
||||
MPI Rank 2: 08/16/2016 10:07:53: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.95254162 * 2560; EvalClassificationError = 0.54179687 * 2560; time = 0.5104s; samplesPerSecond = 5016.0
|
||||
MPI Rank 2: 08/16/2016 10:07:53: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04869166 * 20480; EvalClassificationError = 0.55878906 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.87459s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:07:53: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:07:53: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 10:07:54: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97049696 * 10240; EvalErrorPrediction = 0.54658203 * 10240; time = 1.3532s; samplesPerSecond = 7567.4
|
||||
MPI Rank 2: 08/16/2016 10:07:56: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96995673 * 10240; EvalErrorPrediction = 0.54746094 * 10240; time = 1.6847s; samplesPerSecond = 6078.4
|
||||
MPI Rank 2: 08/16/2016 10:07:56: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.97022685 * 20480; EvalErrorPrediction = 0.54702148 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.05602s
|
||||
MPI Rank 2: 08/16/2016 10:07:54: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97049696 * 10240; EvalClassificationError = 0.54658203 * 10240; time = 1.3532s; samplesPerSecond = 7567.4
|
||||
MPI Rank 2: 08/16/2016 10:07:56: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96995673 * 10240; EvalClassificationError = 0.54746094 * 10240; time = 1.6847s; samplesPerSecond = 6078.4
|
||||
MPI Rank 2: 08/16/2016 10:07:56: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.97022685 * 20480; EvalClassificationError = 0.54702148 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.05602s
|
||||
MPI Rank 2: 08/16/2016 10:07:56: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:07:56: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -133,7 +133,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -159,7 +159,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -239,7 +239,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -265,7 +265,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -351,7 +351,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -377,7 +377,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -474,7 +474,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -503,7 +503,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -526,7 +526,7 @@ MPI Rank 0: 08/16/2016 10:08:02: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 10:08:02: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:02: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 10:08:02: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 10:08:02: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -579,63 +579,63 @@ MPI Rank 0: 08/16/2016 10:08:05: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:05: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1005s; samplesPerSecond = 6368.3
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0993s; samplesPerSecond = 6448.0
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0993s; samplesPerSecond = 6448.3
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0989s; samplesPerSecond = 6468.5
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0980s; samplesPerSecond = 6528.0
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0981s; samplesPerSecond = 6524.2
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6524.2
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.0981s; samplesPerSecond = 6520.9
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0981s; samplesPerSecond = 6526.4
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0981s; samplesPerSecond = 6525.6
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0981s; samplesPerSecond = 6523.4
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0981s; samplesPerSecond = 6523.0
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0981s; samplesPerSecond = 6523.0
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0980s; samplesPerSecond = 6528.5
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0981s; samplesPerSecond = 6524.4
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0981s; samplesPerSecond = 6526.9
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0980s; samplesPerSecond = 6527.3
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0981s; samplesPerSecond = 6524.0
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0983s; samplesPerSecond = 6509.3
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0981s; samplesPerSecond = 6522.4
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0981s; samplesPerSecond = 6521.2
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6525.2
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6525.4
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0980s; samplesPerSecond = 6529.1
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0981s; samplesPerSecond = 6524.0
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0981s; samplesPerSecond = 6524.9
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0981s; samplesPerSecond = 6521.7
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0981s; samplesPerSecond = 6524.1
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0982s; samplesPerSecond = 6520.4
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0981s; samplesPerSecond = 6521.2
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0982s; samplesPerSecond = 6520.4
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0981s; samplesPerSecond = 6521.0
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.14856s
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1005s; samplesPerSecond = 6368.3
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0993s; samplesPerSecond = 6448.0
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0993s; samplesPerSecond = 6448.3
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0989s; samplesPerSecond = 6468.5
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0980s; samplesPerSecond = 6528.0
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0981s; samplesPerSecond = 6524.2
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6524.2
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.0981s; samplesPerSecond = 6520.9
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0981s; samplesPerSecond = 6526.4
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0981s; samplesPerSecond = 6525.6
|
||||
MPI Rank 0: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0981s; samplesPerSecond = 6523.4
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0981s; samplesPerSecond = 6523.0
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0981s; samplesPerSecond = 6523.0
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0980s; samplesPerSecond = 6528.5
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0981s; samplesPerSecond = 6524.4
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0981s; samplesPerSecond = 6526.9
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0980s; samplesPerSecond = 6527.3
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0981s; samplesPerSecond = 6524.0
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0983s; samplesPerSecond = 6509.3
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0981s; samplesPerSecond = 6522.4
|
||||
MPI Rank 0: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0981s; samplesPerSecond = 6521.2
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6525.2
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6525.4
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0980s; samplesPerSecond = 6529.1
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0981s; samplesPerSecond = 6524.0
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0981s; samplesPerSecond = 6524.9
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0981s; samplesPerSecond = 6521.7
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0981s; samplesPerSecond = 6524.1
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0982s; samplesPerSecond = 6520.4
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0981s; samplesPerSecond = 6521.2
|
||||
MPI Rank 0: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0982s; samplesPerSecond = 6520.4
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0981s; samplesPerSecond = 6521.0
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.14856s
|
||||
MPI Rank 0: 08/16/2016 10:08:09: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_Parallel1BitQuantization@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalErrorPrediction = 0.60039062 * 2560; time = 0.1484s; samplesPerSecond = 17245.9
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalErrorPrediction = 0.57070312 * 2560; time = 0.1433s; samplesPerSecond = 17868.6
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalErrorPrediction = 0.56289062 * 2560; time = 0.1419s; samplesPerSecond = 18039.2
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalErrorPrediction = 0.56171875 * 2560; time = 0.1419s; samplesPerSecond = 18035.9
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalErrorPrediction = 0.55976563 * 2560; time = 0.1417s; samplesPerSecond = 18063.5
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalErrorPrediction = 0.54531250 * 2560; time = 0.1426s; samplesPerSecond = 17950.4
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalErrorPrediction = 0.54765625 * 2560; time = 0.1420s; samplesPerSecond = 18034.0
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalErrorPrediction = 0.55507812 * 2560; time = 0.1422s; samplesPerSecond = 18007.8
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalErrorPrediction = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.15008s
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalClassificationError = 0.60039062 * 2560; time = 0.1484s; samplesPerSecond = 17245.9
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalClassificationError = 0.57070312 * 2560; time = 0.1433s; samplesPerSecond = 17868.6
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalClassificationError = 0.56289062 * 2560; time = 0.1419s; samplesPerSecond = 18039.2
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalClassificationError = 0.56171875 * 2560; time = 0.1419s; samplesPerSecond = 18035.9
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalClassificationError = 0.55976563 * 2560; time = 0.1417s; samplesPerSecond = 18063.5
|
||||
MPI Rank 0: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalClassificationError = 0.54531250 * 2560; time = 0.1426s; samplesPerSecond = 17950.4
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalClassificationError = 0.54765625 * 2560; time = 0.1420s; samplesPerSecond = 18034.0
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalClassificationError = 0.55507812 * 2560; time = 0.1422s; samplesPerSecond = 18007.8
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalClassificationError = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.15008s
|
||||
MPI Rank 0: 08/16/2016 10:08:10: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_Parallel1BitQuantization@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalErrorPrediction = 0.53154297 * 10240; time = 0.2875s; samplesPerSecond = 35614.4
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalErrorPrediction = 0.55019531 * 10240; time = 0.2827s; samplesPerSecond = 36224.7
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalErrorPrediction = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.577049s
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalClassificationError = 0.53154297 * 10240; time = 0.2875s; samplesPerSecond = 35614.4
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalClassificationError = 0.55019531 * 10240; time = 0.2827s; samplesPerSecond = 36224.7
|
||||
MPI Rank 0: 08/16/2016 10:08:10: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalClassificationError = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.577049s
|
||||
MPI Rank 0: 08/16/2016 10:08:10: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_Parallel1BitQuantization@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 10:08:10: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -688,7 +688,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -714,7 +714,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -794,7 +794,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -820,7 +820,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -906,7 +906,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -932,7 +932,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1029,7 +1029,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1058,7 +1058,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1081,7 +1081,7 @@ MPI Rank 1: 08/16/2016 10:08:01: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 10:08:01: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:01: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 10:08:01: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 10:08:01: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1134,61 +1134,61 @@ MPI Rank 1: 08/16/2016 10:08:05: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:05: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1004s; samplesPerSecond = 6376.2
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0993s; samplesPerSecond = 6447.8
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0993s; samplesPerSecond = 6448.2
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0985s; samplesPerSecond = 6499.4
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0973s; samplesPerSecond = 6575.4
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0981s; samplesPerSecond = 6523.8
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6524.0
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.0981s; samplesPerSecond = 6521.2
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0981s; samplesPerSecond = 6526.9
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0981s; samplesPerSecond = 6526.9
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0981s; samplesPerSecond = 6522.8
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0981s; samplesPerSecond = 6522.2
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0981s; samplesPerSecond = 6522.1
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0980s; samplesPerSecond = 6527.8
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0981s; samplesPerSecond = 6524.4
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0981s; samplesPerSecond = 6526.4
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0980s; samplesPerSecond = 6527.4
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0981s; samplesPerSecond = 6524.1
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0983s; samplesPerSecond = 6509.2
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0981s; samplesPerSecond = 6522.3
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0981s; samplesPerSecond = 6521.6
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6523.1
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6525.5
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0980s; samplesPerSecond = 6528.5
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0981s; samplesPerSecond = 6523.4
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0981s; samplesPerSecond = 6524.1
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0982s; samplesPerSecond = 6520.3
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0981s; samplesPerSecond = 6524.7
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0982s; samplesPerSecond = 6519.8
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0982s; samplesPerSecond = 6520.1
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0982s; samplesPerSecond = 6520.1
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0982s; samplesPerSecond = 6519.8
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.14736s
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1004s; samplesPerSecond = 6376.2
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0993s; samplesPerSecond = 6447.8
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0993s; samplesPerSecond = 6448.2
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0985s; samplesPerSecond = 6499.4
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0973s; samplesPerSecond = 6575.4
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0981s; samplesPerSecond = 6523.8
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6524.0
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.0981s; samplesPerSecond = 6521.2
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0981s; samplesPerSecond = 6526.9
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0981s; samplesPerSecond = 6526.9
|
||||
MPI Rank 1: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0981s; samplesPerSecond = 6522.8
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0981s; samplesPerSecond = 6522.2
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0981s; samplesPerSecond = 6522.1
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0980s; samplesPerSecond = 6527.8
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0981s; samplesPerSecond = 6524.4
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0981s; samplesPerSecond = 6526.4
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0980s; samplesPerSecond = 6527.4
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0981s; samplesPerSecond = 6524.1
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0983s; samplesPerSecond = 6509.2
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0981s; samplesPerSecond = 6522.3
|
||||
MPI Rank 1: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0981s; samplesPerSecond = 6521.6
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6523.1
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6525.5
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0980s; samplesPerSecond = 6528.5
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0981s; samplesPerSecond = 6523.4
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0981s; samplesPerSecond = 6524.1
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0982s; samplesPerSecond = 6520.3
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0981s; samplesPerSecond = 6524.7
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0982s; samplesPerSecond = 6519.8
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0982s; samplesPerSecond = 6520.1
|
||||
MPI Rank 1: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0982s; samplesPerSecond = 6520.1
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0982s; samplesPerSecond = 6519.8
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.14736s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalErrorPrediction = 0.60039062 * 2560; time = 0.1487s; samplesPerSecond = 17211.0
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalErrorPrediction = 0.57070312 * 2560; time = 0.1433s; samplesPerSecond = 17870.0
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalErrorPrediction = 0.56289062 * 2560; time = 0.1419s; samplesPerSecond = 18038.8
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalErrorPrediction = 0.56171875 * 2560; time = 0.1419s; samplesPerSecond = 18035.5
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalErrorPrediction = 0.55976563 * 2560; time = 0.1418s; samplesPerSecond = 18060.0
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalErrorPrediction = 0.54531250 * 2560; time = 0.1425s; samplesPerSecond = 17959.2
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalErrorPrediction = 0.54765625 * 2560; time = 0.1420s; samplesPerSecond = 18029.3
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalErrorPrediction = 0.55507812 * 2560; time = 0.1422s; samplesPerSecond = 18004.8
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalErrorPrediction = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.15002s
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalClassificationError = 0.60039062 * 2560; time = 0.1487s; samplesPerSecond = 17211.0
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalClassificationError = 0.57070312 * 2560; time = 0.1433s; samplesPerSecond = 17870.0
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalClassificationError = 0.56289062 * 2560; time = 0.1419s; samplesPerSecond = 18038.8
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalClassificationError = 0.56171875 * 2560; time = 0.1419s; samplesPerSecond = 18035.5
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalClassificationError = 0.55976563 * 2560; time = 0.1418s; samplesPerSecond = 18060.0
|
||||
MPI Rank 1: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalClassificationError = 0.54531250 * 2560; time = 0.1425s; samplesPerSecond = 17959.2
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalClassificationError = 0.54765625 * 2560; time = 0.1420s; samplesPerSecond = 18029.3
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalClassificationError = 0.55507812 * 2560; time = 0.1422s; samplesPerSecond = 18004.8
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalClassificationError = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.15002s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalErrorPrediction = 0.53154297 * 10240; time = 0.2878s; samplesPerSecond = 35576.6
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalErrorPrediction = 0.55019531 * 10240; time = 0.2827s; samplesPerSecond = 36218.6
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalErrorPrediction = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.576897s
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalClassificationError = 0.53154297 * 10240; time = 0.2878s; samplesPerSecond = 35576.6
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalClassificationError = 0.55019531 * 10240; time = 0.2827s; samplesPerSecond = 36218.6
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalClassificationError = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.576897s
|
||||
MPI Rank 1: 08/16/2016 10:08:10: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:10: Action "train" complete.
|
||||
|
|
@ -1240,7 +1240,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1266,7 +1266,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1346,7 +1346,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1372,7 +1372,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1458,7 +1458,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1484,7 +1484,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1581,7 +1581,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1610,7 +1610,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1633,7 +1633,7 @@ MPI Rank 2: 08/16/2016 10:08:03: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 10:08:03: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:08:03: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 10:08:03: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 10:08:03: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1686,61 +1686,61 @@ MPI Rank 2: 08/16/2016 10:08:05: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:08:05: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1006s; samplesPerSecond = 6362.3
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0993s; samplesPerSecond = 6446.8
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0992s; samplesPerSecond = 6448.4
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0989s; samplesPerSecond = 6470.1
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0981s; samplesPerSecond = 6527.1
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0981s; samplesPerSecond = 6524.4
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6524.2
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.0981s; samplesPerSecond = 6521.2
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0981s; samplesPerSecond = 6526.5
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0981s; samplesPerSecond = 6525.6
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0981s; samplesPerSecond = 6523.6
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0981s; samplesPerSecond = 6522.4
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0981s; samplesPerSecond = 6523.6
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0980s; samplesPerSecond = 6528.4
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0981s; samplesPerSecond = 6524.6
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0981s; samplesPerSecond = 6527.0
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0980s; samplesPerSecond = 6527.3
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0981s; samplesPerSecond = 6524.6
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0983s; samplesPerSecond = 6509.5
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0981s; samplesPerSecond = 6522.4
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0981s; samplesPerSecond = 6521.4
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6525.3
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6525.5
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0980s; samplesPerSecond = 6529.3
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0981s; samplesPerSecond = 6524.0
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0981s; samplesPerSecond = 6524.9
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0981s; samplesPerSecond = 6521.8
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0981s; samplesPerSecond = 6524.2
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0982s; samplesPerSecond = 6519.7
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0981s; samplesPerSecond = 6521.7
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0981s; samplesPerSecond = 6520.7
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0982s; samplesPerSecond = 6520.3
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.14859s
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1006s; samplesPerSecond = 6362.3
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0993s; samplesPerSecond = 6446.8
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0992s; samplesPerSecond = 6448.4
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0989s; samplesPerSecond = 6470.1
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0981s; samplesPerSecond = 6527.1
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0981s; samplesPerSecond = 6524.4
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6524.2
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.0981s; samplesPerSecond = 6521.2
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0981s; samplesPerSecond = 6526.5
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0981s; samplesPerSecond = 6525.6
|
||||
MPI Rank 2: 08/16/2016 10:08:06: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0981s; samplesPerSecond = 6523.6
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0981s; samplesPerSecond = 6522.4
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0981s; samplesPerSecond = 6523.6
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0980s; samplesPerSecond = 6528.4
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0981s; samplesPerSecond = 6524.6
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0981s; samplesPerSecond = 6527.0
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0980s; samplesPerSecond = 6527.3
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0981s; samplesPerSecond = 6524.6
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0983s; samplesPerSecond = 6509.5
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0981s; samplesPerSecond = 6522.4
|
||||
MPI Rank 2: 08/16/2016 10:08:07: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0981s; samplesPerSecond = 6521.4
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6525.3
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0981s; samplesPerSecond = 6525.5
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0980s; samplesPerSecond = 6529.3
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0981s; samplesPerSecond = 6524.0
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0981s; samplesPerSecond = 6524.9
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0981s; samplesPerSecond = 6521.8
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0981s; samplesPerSecond = 6524.2
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0982s; samplesPerSecond = 6519.7
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0981s; samplesPerSecond = 6521.7
|
||||
MPI Rank 2: 08/16/2016 10:08:08: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0981s; samplesPerSecond = 6520.7
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0982s; samplesPerSecond = 6520.3
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.14859s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalErrorPrediction = 0.60039062 * 2560; time = 0.1485s; samplesPerSecond = 17237.9
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalErrorPrediction = 0.57070312 * 2560; time = 0.1432s; samplesPerSecond = 17878.8
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalErrorPrediction = 0.56289062 * 2560; time = 0.1419s; samplesPerSecond = 18040.0
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalErrorPrediction = 0.56171875 * 2560; time = 0.1419s; samplesPerSecond = 18036.3
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalErrorPrediction = 0.55976563 * 2560; time = 0.1420s; samplesPerSecond = 18024.4
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalErrorPrediction = 0.54531250 * 2560; time = 0.1427s; samplesPerSecond = 17945.9
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalErrorPrediction = 0.54765625 * 2560; time = 0.1420s; samplesPerSecond = 18031.1
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalErrorPrediction = 0.55507812 * 2560; time = 0.1422s; samplesPerSecond = 18006.9
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalErrorPrediction = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.14989s
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalClassificationError = 0.60039062 * 2560; time = 0.1485s; samplesPerSecond = 17237.9
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalClassificationError = 0.57070312 * 2560; time = 0.1432s; samplesPerSecond = 17878.8
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalClassificationError = 0.56289062 * 2560; time = 0.1419s; samplesPerSecond = 18040.0
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalClassificationError = 0.56171875 * 2560; time = 0.1419s; samplesPerSecond = 18036.3
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalClassificationError = 0.55976563 * 2560; time = 0.1420s; samplesPerSecond = 18024.4
|
||||
MPI Rank 2: 08/16/2016 10:08:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalClassificationError = 0.54531250 * 2560; time = 0.1427s; samplesPerSecond = 17945.9
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalClassificationError = 0.54765625 * 2560; time = 0.1420s; samplesPerSecond = 18031.1
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalClassificationError = 0.55507812 * 2560; time = 0.1422s; samplesPerSecond = 18006.9
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalClassificationError = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.14989s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalErrorPrediction = 0.53154297 * 10240; time = 0.2880s; samplesPerSecond = 35554.3
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalErrorPrediction = 0.55019531 * 10240; time = 0.2824s; samplesPerSecond = 36264.2
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalErrorPrediction = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.57713s
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalClassificationError = 0.53154297 * 10240; time = 0.2880s; samplesPerSecond = 35554.3
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalClassificationError = 0.55019531 * 10240; time = 0.2824s; samplesPerSecond = 36264.2
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalClassificationError = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.57713s
|
||||
MPI Rank 2: 08/16/2016 10:08:10: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:08:10: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -131,7 +131,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -157,7 +157,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -237,7 +237,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -263,7 +263,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -349,7 +349,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -375,7 +375,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -471,7 +471,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -500,7 +500,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -523,7 +523,7 @@ MPI Rank 0: 08/16/2016 03:19:00: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:19:00: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:00: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:19:00: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:19:00: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -576,63 +576,63 @@ MPI Rank 0: 08/16/2016 03:19:04: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.1132s; samplesPerSecond = 5655.1
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.0982s; samplesPerSecond = 6516.7
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.0988s; samplesPerSecond = 6478.8
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1000s; samplesPerSecond = 6398.8
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.1040s; samplesPerSecond = 6154.4
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.1033s; samplesPerSecond = 6193.7
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.1016s; samplesPerSecond = 6297.0
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.1058s; samplesPerSecond = 6050.7
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.1060s; samplesPerSecond = 6035.6
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.1048s; samplesPerSecond = 6108.5
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.0996s; samplesPerSecond = 6425.8
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.1049s; samplesPerSecond = 6101.1
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.1018s; samplesPerSecond = 6287.6
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.1019s; samplesPerSecond = 6280.1
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.1007s; samplesPerSecond = 6356.0
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.1023s; samplesPerSecond = 6253.3
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.1118s; samplesPerSecond = 5722.9
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.1048s; samplesPerSecond = 6106.2
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.1044s; samplesPerSecond = 6132.4
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.1032s; samplesPerSecond = 6199.4
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.1020s; samplesPerSecond = 6273.4
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.1006s; samplesPerSecond = 6363.6
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.1043s; samplesPerSecond = 6134.7
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1053s; samplesPerSecond = 6076.7
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.1004s; samplesPerSecond = 6372.9
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.1007s; samplesPerSecond = 6354.1
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.1028s; samplesPerSecond = 6224.7
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1003s; samplesPerSecond = 6380.2
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0934s; samplesPerSecond = 6852.8
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0901s; samplesPerSecond = 7103.6
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0900s; samplesPerSecond = 7109.4
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.0885s; samplesPerSecond = 7231.4
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.27512s
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.1132s; samplesPerSecond = 5655.1
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.0982s; samplesPerSecond = 6516.7
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.0988s; samplesPerSecond = 6478.8
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1000s; samplesPerSecond = 6398.8
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.1040s; samplesPerSecond = 6154.4
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.1033s; samplesPerSecond = 6193.7
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.1016s; samplesPerSecond = 6297.0
|
||||
MPI Rank 0: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.1058s; samplesPerSecond = 6050.7
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.1060s; samplesPerSecond = 6035.6
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.1048s; samplesPerSecond = 6108.5
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.0996s; samplesPerSecond = 6425.8
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.1049s; samplesPerSecond = 6101.1
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.1018s; samplesPerSecond = 6287.6
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.1019s; samplesPerSecond = 6280.1
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.1007s; samplesPerSecond = 6356.0
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.1023s; samplesPerSecond = 6253.3
|
||||
MPI Rank 0: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.1118s; samplesPerSecond = 5722.9
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.1048s; samplesPerSecond = 6106.2
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.1044s; samplesPerSecond = 6132.4
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.1032s; samplesPerSecond = 6199.4
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.1020s; samplesPerSecond = 6273.4
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.1006s; samplesPerSecond = 6363.6
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.1043s; samplesPerSecond = 6134.7
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1053s; samplesPerSecond = 6076.7
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.1004s; samplesPerSecond = 6372.9
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.1007s; samplesPerSecond = 6354.1
|
||||
MPI Rank 0: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.1028s; samplesPerSecond = 6224.7
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1003s; samplesPerSecond = 6380.2
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0934s; samplesPerSecond = 6852.8
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0901s; samplesPerSecond = 7103.6
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0900s; samplesPerSecond = 7109.4
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.0885s; samplesPerSecond = 7231.4
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.27512s
|
||||
MPI Rank 0: 08/16/2016 03:19:07: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_Parallel1BitQuantization@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.15042048 * 2560; EvalErrorPrediction = 0.58867187 * 2560; time = 0.3458s; samplesPerSecond = 7402.6
|
||||
MPI Rank 0: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.11694314 * 2560; EvalErrorPrediction = 0.56132812 * 2560; time = 0.3452s; samplesPerSecond = 7416.2
|
||||
MPI Rank 0: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.05986597 * 2560; EvalErrorPrediction = 0.56523437 * 2560; time = 0.3527s; samplesPerSecond = 7258.5
|
||||
MPI Rank 0: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.02348440 * 2560; EvalErrorPrediction = 0.55820313 * 2560; time = 0.3475s; samplesPerSecond = 7366.4
|
||||
MPI Rank 0: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02257161 * 2560; EvalErrorPrediction = 0.54921875 * 2560; time = 0.3430s; samplesPerSecond = 7463.0
|
||||
MPI Rank 0: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96745389 * 2560; EvalErrorPrediction = 0.53867188 * 2560; time = 0.3327s; samplesPerSecond = 7694.1
|
||||
MPI Rank 0: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95891510 * 2560; EvalErrorPrediction = 0.52851563 * 2560; time = 0.3305s; samplesPerSecond = 7745.6
|
||||
MPI Rank 0: 08/16/2016 03:19:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.93032448 * 2560; EvalErrorPrediction = 0.53281250 * 2560; time = 0.3646s; samplesPerSecond = 7021.8
|
||||
MPI Rank 0: 08/16/2016 03:19:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02874738 * 20480; EvalErrorPrediction = 0.55283203 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=2.79211s
|
||||
MPI Rank 0: 08/16/2016 03:19:07: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.15042048 * 2560; EvalClassificationError = 0.58867187 * 2560; time = 0.3458s; samplesPerSecond = 7402.6
|
||||
MPI Rank 0: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.11694314 * 2560; EvalClassificationError = 0.56132812 * 2560; time = 0.3452s; samplesPerSecond = 7416.2
|
||||
MPI Rank 0: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.05986597 * 2560; EvalClassificationError = 0.56523437 * 2560; time = 0.3527s; samplesPerSecond = 7258.5
|
||||
MPI Rank 0: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.02348440 * 2560; EvalClassificationError = 0.55820313 * 2560; time = 0.3475s; samplesPerSecond = 7366.4
|
||||
MPI Rank 0: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02257161 * 2560; EvalClassificationError = 0.54921875 * 2560; time = 0.3430s; samplesPerSecond = 7463.0
|
||||
MPI Rank 0: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96745389 * 2560; EvalClassificationError = 0.53867188 * 2560; time = 0.3327s; samplesPerSecond = 7694.1
|
||||
MPI Rank 0: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95891510 * 2560; EvalClassificationError = 0.52851563 * 2560; time = 0.3305s; samplesPerSecond = 7745.6
|
||||
MPI Rank 0: 08/16/2016 03:19:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.93032448 * 2560; EvalClassificationError = 0.53281250 * 2560; time = 0.3646s; samplesPerSecond = 7021.8
|
||||
MPI Rank 0: 08/16/2016 03:19:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02874738 * 20480; EvalClassificationError = 0.55283203 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=2.79211s
|
||||
MPI Rank 0: 08/16/2016 03:19:10: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_Parallel1BitQuantization@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:10: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:10: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95622782 * 10240; EvalErrorPrediction = 0.54609375 * 10240; time = 0.6987s; samplesPerSecond = 14654.8
|
||||
MPI Rank 0: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95081435 * 10240; EvalErrorPrediction = 0.54355469 * 10240; time = 0.6563s; samplesPerSecond = 15601.8
|
||||
MPI Rank 0: 08/16/2016 03:19:11: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95352108 * 20480; EvalErrorPrediction = 0.54482422 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.38257s
|
||||
MPI Rank 0: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95622782 * 10240; EvalClassificationError = 0.54609375 * 10240; time = 0.6987s; samplesPerSecond = 14654.8
|
||||
MPI Rank 0: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95081435 * 10240; EvalClassificationError = 0.54355469 * 10240; time = 0.6563s; samplesPerSecond = 15601.8
|
||||
MPI Rank 0: 08/16/2016 03:19:11: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95352108 * 20480; EvalClassificationError = 0.54482422 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.38257s
|
||||
MPI Rank 0: 08/16/2016 03:19:11: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_Parallel1BitQuantization@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:19:12: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -686,7 +686,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -712,7 +712,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -792,7 +792,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -818,7 +818,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -904,7 +904,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -930,7 +930,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1026,7 +1026,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1055,7 +1055,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1078,7 +1078,7 @@ MPI Rank 1: 08/16/2016 03:19:00: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:19:00: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:00: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:19:00: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:19:00: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1131,61 +1131,61 @@ MPI Rank 1: 08/16/2016 03:19:04: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.1184s; samplesPerSecond = 5404.2
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.0985s; samplesPerSecond = 6499.5
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.1028s; samplesPerSecond = 6226.7
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0998s; samplesPerSecond = 6412.8
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0983s; samplesPerSecond = 6513.6
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.0977s; samplesPerSecond = 6550.6
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.1028s; samplesPerSecond = 6224.5
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.0998s; samplesPerSecond = 6411.4
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.1033s; samplesPerSecond = 6193.1
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.1021s; samplesPerSecond = 6270.7
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.0947s; samplesPerSecond = 6757.0
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0993s; samplesPerSecond = 6443.8
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.1026s; samplesPerSecond = 6236.5
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.1053s; samplesPerSecond = 6079.8
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.1079s; samplesPerSecond = 5929.4
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.1038s; samplesPerSecond = 6167.4
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.1006s; samplesPerSecond = 6362.9
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.1004s; samplesPerSecond = 6372.7
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.0987s; samplesPerSecond = 6481.9
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.0996s; samplesPerSecond = 6428.0
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.1059s; samplesPerSecond = 6041.7
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.1059s; samplesPerSecond = 6044.4
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.1055s; samplesPerSecond = 6067.4
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1031s; samplesPerSecond = 6207.6
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.1068s; samplesPerSecond = 5994.6
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.1001s; samplesPerSecond = 6392.7
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.1046s; samplesPerSecond = 6118.6
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1057s; samplesPerSecond = 6053.9
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1048s; samplesPerSecond = 6105.4
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0922s; samplesPerSecond = 6938.0
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0938s; samplesPerSecond = 6824.1
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.0847s; samplesPerSecond = 7551.9
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.27584s
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.1184s; samplesPerSecond = 5404.2
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.0985s; samplesPerSecond = 6499.5
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.1028s; samplesPerSecond = 6226.7
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0998s; samplesPerSecond = 6412.8
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0983s; samplesPerSecond = 6513.6
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.0977s; samplesPerSecond = 6550.6
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.1028s; samplesPerSecond = 6224.5
|
||||
MPI Rank 1: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.0998s; samplesPerSecond = 6411.4
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.1033s; samplesPerSecond = 6193.1
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.1021s; samplesPerSecond = 6270.7
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.0947s; samplesPerSecond = 6757.0
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0993s; samplesPerSecond = 6443.8
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.1026s; samplesPerSecond = 6236.5
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.1053s; samplesPerSecond = 6079.8
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.1079s; samplesPerSecond = 5929.4
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.1038s; samplesPerSecond = 6167.4
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.1006s; samplesPerSecond = 6362.9
|
||||
MPI Rank 1: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.1004s; samplesPerSecond = 6372.7
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.0987s; samplesPerSecond = 6481.9
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.0996s; samplesPerSecond = 6428.0
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.1059s; samplesPerSecond = 6041.7
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.1059s; samplesPerSecond = 6044.4
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.1055s; samplesPerSecond = 6067.4
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1031s; samplesPerSecond = 6207.6
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.1068s; samplesPerSecond = 5994.6
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.1001s; samplesPerSecond = 6392.7
|
||||
MPI Rank 1: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.1046s; samplesPerSecond = 6118.6
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1057s; samplesPerSecond = 6053.9
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1048s; samplesPerSecond = 6105.4
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0922s; samplesPerSecond = 6938.0
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0938s; samplesPerSecond = 6824.1
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.0847s; samplesPerSecond = 7551.9
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.27584s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.15042048 * 2560; EvalErrorPrediction = 0.58867187 * 2560; time = 0.3464s; samplesPerSecond = 7390.3
|
||||
MPI Rank 1: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.11694314 * 2560; EvalErrorPrediction = 0.56132812 * 2560; time = 0.3453s; samplesPerSecond = 7413.0
|
||||
MPI Rank 1: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.05986597 * 2560; EvalErrorPrediction = 0.56523437 * 2560; time = 0.3528s; samplesPerSecond = 7255.6
|
||||
MPI Rank 1: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.02348440 * 2560; EvalErrorPrediction = 0.55820313 * 2560; time = 0.3493s; samplesPerSecond = 7329.5
|
||||
MPI Rank 1: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02257161 * 2560; EvalErrorPrediction = 0.54921875 * 2560; time = 0.3430s; samplesPerSecond = 7464.6
|
||||
MPI Rank 1: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96745389 * 2560; EvalErrorPrediction = 0.53867188 * 2560; time = 0.3326s; samplesPerSecond = 7696.2
|
||||
MPI Rank 1: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95891510 * 2560; EvalErrorPrediction = 0.52851563 * 2560; time = 0.3306s; samplesPerSecond = 7743.2
|
||||
MPI Rank 1: 08/16/2016 03:19:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.93032448 * 2560; EvalErrorPrediction = 0.53281250 * 2560; time = 0.3643s; samplesPerSecond = 7026.8
|
||||
MPI Rank 1: 08/16/2016 03:19:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02874738 * 20480; EvalErrorPrediction = 0.55283203 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=2.79338s
|
||||
MPI Rank 1: 08/16/2016 03:19:07: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.15042048 * 2560; EvalClassificationError = 0.58867187 * 2560; time = 0.3464s; samplesPerSecond = 7390.3
|
||||
MPI Rank 1: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.11694314 * 2560; EvalClassificationError = 0.56132812 * 2560; time = 0.3453s; samplesPerSecond = 7413.0
|
||||
MPI Rank 1: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.05986597 * 2560; EvalClassificationError = 0.56523437 * 2560; time = 0.3528s; samplesPerSecond = 7255.6
|
||||
MPI Rank 1: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.02348440 * 2560; EvalClassificationError = 0.55820313 * 2560; time = 0.3493s; samplesPerSecond = 7329.5
|
||||
MPI Rank 1: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02257161 * 2560; EvalClassificationError = 0.54921875 * 2560; time = 0.3430s; samplesPerSecond = 7464.6
|
||||
MPI Rank 1: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96745389 * 2560; EvalClassificationError = 0.53867188 * 2560; time = 0.3326s; samplesPerSecond = 7696.2
|
||||
MPI Rank 1: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95891510 * 2560; EvalClassificationError = 0.52851563 * 2560; time = 0.3306s; samplesPerSecond = 7743.2
|
||||
MPI Rank 1: 08/16/2016 03:19:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.93032448 * 2560; EvalClassificationError = 0.53281250 * 2560; time = 0.3643s; samplesPerSecond = 7026.8
|
||||
MPI Rank 1: 08/16/2016 03:19:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02874738 * 20480; EvalClassificationError = 0.55283203 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=2.79338s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:10: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:10: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95622782 * 10240; EvalErrorPrediction = 0.54609375 * 10240; time = 0.7010s; samplesPerSecond = 14607.7
|
||||
MPI Rank 1: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95081435 * 10240; EvalErrorPrediction = 0.54355469 * 10240; time = 0.6562s; samplesPerSecond = 15604.6
|
||||
MPI Rank 1: 08/16/2016 03:19:11: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95352108 * 20480; EvalErrorPrediction = 0.54482422 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.38447s
|
||||
MPI Rank 1: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95622782 * 10240; EvalClassificationError = 0.54609375 * 10240; time = 0.7010s; samplesPerSecond = 14607.7
|
||||
MPI Rank 1: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95081435 * 10240; EvalClassificationError = 0.54355469 * 10240; time = 0.6562s; samplesPerSecond = 15604.6
|
||||
MPI Rank 1: 08/16/2016 03:19:11: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95352108 * 20480; EvalClassificationError = 0.54482422 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.38447s
|
||||
MPI Rank 1: 08/16/2016 03:19:12: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:12: Action "train" complete.
|
||||
|
|
@ -1238,7 +1238,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1264,7 +1264,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1344,7 +1344,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1370,7 +1370,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1456,7 +1456,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1482,7 +1482,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1578,7 +1578,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1607,7 +1607,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1630,7 +1630,7 @@ MPI Rank 2: 08/16/2016 03:19:00: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 03:19:00: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:00: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 03:19:00: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 03:19:00: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1683,61 +1683,61 @@ MPI Rank 2: 08/16/2016 03:19:04: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.1012s; samplesPerSecond = 6326.6
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.0866s; samplesPerSecond = 7392.5
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.0904s; samplesPerSecond = 7082.9
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0912s; samplesPerSecond = 7021.0
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0943s; samplesPerSecond = 6785.6
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.0933s; samplesPerSecond = 6862.1
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.1051s; samplesPerSecond = 6091.7
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.0906s; samplesPerSecond = 7060.9
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.0886s; samplesPerSecond = 7223.9
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.0905s; samplesPerSecond = 7074.5
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.0904s; samplesPerSecond = 7079.9
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0886s; samplesPerSecond = 7222.9
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.0870s; samplesPerSecond = 7353.4
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.0839s; samplesPerSecond = 7626.0
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.0872s; samplesPerSecond = 7338.4
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.0913s; samplesPerSecond = 7012.1
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.0937s; samplesPerSecond = 6833.4
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.0912s; samplesPerSecond = 7016.6
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.0892s; samplesPerSecond = 7177.7
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.0879s; samplesPerSecond = 7276.9
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.0916s; samplesPerSecond = 6990.4
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.0886s; samplesPerSecond = 7222.5
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0929s; samplesPerSecond = 6890.6
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0925s; samplesPerSecond = 6917.7
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.0935s; samplesPerSecond = 6841.6
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.1001s; samplesPerSecond = 6395.3
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0952s; samplesPerSecond = 6724.1
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0954s; samplesPerSecond = 6706.4
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0891s; samplesPerSecond = 7182.8
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0879s; samplesPerSecond = 7281.7
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0911s; samplesPerSecond = 7021.8
|
||||
MPI Rank 2: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.0904s; samplesPerSecond = 7076.3
|
||||
MPI Rank 2: 08/16/2016 03:19:07: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.95598s
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.1012s; samplesPerSecond = 6326.6
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.0866s; samplesPerSecond = 7392.5
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.0904s; samplesPerSecond = 7082.9
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0912s; samplesPerSecond = 7021.0
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0943s; samplesPerSecond = 6785.6
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.0933s; samplesPerSecond = 6862.1
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.1051s; samplesPerSecond = 6091.7
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.0906s; samplesPerSecond = 7060.9
|
||||
MPI Rank 2: 08/16/2016 03:19:04: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.0886s; samplesPerSecond = 7223.9
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.0905s; samplesPerSecond = 7074.5
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.0904s; samplesPerSecond = 7079.9
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0886s; samplesPerSecond = 7222.9
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.0870s; samplesPerSecond = 7353.4
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.0839s; samplesPerSecond = 7626.0
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.0872s; samplesPerSecond = 7338.4
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.0913s; samplesPerSecond = 7012.1
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.0937s; samplesPerSecond = 6833.4
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.0912s; samplesPerSecond = 7016.6
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.0892s; samplesPerSecond = 7177.7
|
||||
MPI Rank 2: 08/16/2016 03:19:05: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.0879s; samplesPerSecond = 7276.9
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.0916s; samplesPerSecond = 6990.4
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.0886s; samplesPerSecond = 7222.5
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0929s; samplesPerSecond = 6890.6
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0925s; samplesPerSecond = 6917.7
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.0935s; samplesPerSecond = 6841.6
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.1001s; samplesPerSecond = 6395.3
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0952s; samplesPerSecond = 6724.1
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0954s; samplesPerSecond = 6706.4
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0891s; samplesPerSecond = 7182.8
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0879s; samplesPerSecond = 7281.7
|
||||
MPI Rank 2: 08/16/2016 03:19:06: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0911s; samplesPerSecond = 7021.8
|
||||
MPI Rank 2: 08/16/2016 03:19:07: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.0904s; samplesPerSecond = 7076.3
|
||||
MPI Rank 2: 08/16/2016 03:19:07: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.95598s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:07: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:07: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 03:19:07: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.15042048 * 2560; EvalErrorPrediction = 0.58867187 * 2560; time = 0.3462s; samplesPerSecond = 7394.9
|
||||
MPI Rank 2: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.11694314 * 2560; EvalErrorPrediction = 0.56132812 * 2560; time = 0.3453s; samplesPerSecond = 7413.9
|
||||
MPI Rank 2: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.05986597 * 2560; EvalErrorPrediction = 0.56523437 * 2560; time = 0.3527s; samplesPerSecond = 7258.6
|
||||
MPI Rank 2: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.02348440 * 2560; EvalErrorPrediction = 0.55820313 * 2560; time = 0.3493s; samplesPerSecond = 7328.9
|
||||
MPI Rank 2: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02257161 * 2560; EvalErrorPrediction = 0.54921875 * 2560; time = 0.3429s; samplesPerSecond = 7466.5
|
||||
MPI Rank 2: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96745389 * 2560; EvalErrorPrediction = 0.53867188 * 2560; time = 0.3327s; samplesPerSecond = 7695.2
|
||||
MPI Rank 2: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95891510 * 2560; EvalErrorPrediction = 0.52851563 * 2560; time = 0.3305s; samplesPerSecond = 7745.5
|
||||
MPI Rank 2: 08/16/2016 03:19:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.93032448 * 2560; EvalErrorPrediction = 0.53281250 * 2560; time = 0.3646s; samplesPerSecond = 7021.5
|
||||
MPI Rank 2: 08/16/2016 03:19:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02874738 * 20480; EvalErrorPrediction = 0.55283203 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=2.79383s
|
||||
MPI Rank 2: 08/16/2016 03:19:07: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.15042048 * 2560; EvalClassificationError = 0.58867187 * 2560; time = 0.3462s; samplesPerSecond = 7394.9
|
||||
MPI Rank 2: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.11694314 * 2560; EvalClassificationError = 0.56132812 * 2560; time = 0.3453s; samplesPerSecond = 7413.9
|
||||
MPI Rank 2: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.05986597 * 2560; EvalClassificationError = 0.56523437 * 2560; time = 0.3527s; samplesPerSecond = 7258.6
|
||||
MPI Rank 2: 08/16/2016 03:19:08: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.02348440 * 2560; EvalClassificationError = 0.55820313 * 2560; time = 0.3493s; samplesPerSecond = 7328.9
|
||||
MPI Rank 2: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02257161 * 2560; EvalClassificationError = 0.54921875 * 2560; time = 0.3429s; samplesPerSecond = 7466.5
|
||||
MPI Rank 2: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96745389 * 2560; EvalClassificationError = 0.53867188 * 2560; time = 0.3327s; samplesPerSecond = 7695.2
|
||||
MPI Rank 2: 08/16/2016 03:19:09: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95891510 * 2560; EvalClassificationError = 0.52851563 * 2560; time = 0.3305s; samplesPerSecond = 7745.5
|
||||
MPI Rank 2: 08/16/2016 03:19:10: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.93032448 * 2560; EvalClassificationError = 0.53281250 * 2560; time = 0.3646s; samplesPerSecond = 7021.5
|
||||
MPI Rank 2: 08/16/2016 03:19:10: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02874738 * 20480; EvalClassificationError = 0.55283203 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=2.79383s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:10: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:10: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95622782 * 10240; EvalErrorPrediction = 0.54609375 * 10240; time = 0.7014s; samplesPerSecond = 14599.6
|
||||
MPI Rank 2: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95081435 * 10240; EvalErrorPrediction = 0.54355469 * 10240; time = 0.6564s; samplesPerSecond = 15599.2
|
||||
MPI Rank 2: 08/16/2016 03:19:11: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95352108 * 20480; EvalErrorPrediction = 0.54482422 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.38407s
|
||||
MPI Rank 2: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95622782 * 10240; EvalClassificationError = 0.54609375 * 10240; time = 0.7014s; samplesPerSecond = 14599.6
|
||||
MPI Rank 2: 08/16/2016 03:19:11: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95081435 * 10240; EvalClassificationError = 0.54355469 * 10240; time = 0.6564s; samplesPerSecond = 15599.2
|
||||
MPI Rank 2: 08/16/2016 03:19:11: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95352108 * 20480; EvalClassificationError = 0.54482422 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.38407s
|
||||
MPI Rank 2: 08/16/2016 03:19:12: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:12: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -131,7 +131,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -157,7 +157,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -237,7 +237,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -263,7 +263,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -349,7 +349,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -375,7 +375,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -472,7 +472,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -501,7 +501,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -524,7 +524,7 @@ MPI Rank 0: 08/16/2016 03:19:22: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:19:22: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:22: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:19:22: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:19:22: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -577,63 +577,63 @@ MPI Rank 0: 08/16/2016 03:19:28: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:28: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.0934s; samplesPerSecond = 6851.1
|
||||
MPI Rank 0: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0945s; samplesPerSecond = 6775.2
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.1008s; samplesPerSecond = 6351.3
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0851s; samplesPerSecond = 7522.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.1101s; samplesPerSecond = 5814.1
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0879s; samplesPerSecond = 7280.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0876s; samplesPerSecond = 7306.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.1010s; samplesPerSecond = 6338.0
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0915s; samplesPerSecond = 6997.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.1070s; samplesPerSecond = 5978.9
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0879s; samplesPerSecond = 7277.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0881s; samplesPerSecond = 7264.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0948s; samplesPerSecond = 6750.9
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0913s; samplesPerSecond = 7013.3
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0928s; samplesPerSecond = 6897.2
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.1049s; samplesPerSecond = 6100.2
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0865s; samplesPerSecond = 7398.0
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0889s; samplesPerSecond = 7202.8
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0971s; samplesPerSecond = 6594.3
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0983s; samplesPerSecond = 6511.5
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1010s; samplesPerSecond = 6339.3
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0944s; samplesPerSecond = 6782.8
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1063s; samplesPerSecond = 6018.7
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1045s; samplesPerSecond = 6124.6
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0902s; samplesPerSecond = 7096.6
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.1010s; samplesPerSecond = 6336.6
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0913s; samplesPerSecond = 7013.5
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0944s; samplesPerSecond = 6776.6
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0897s; samplesPerSecond = 7136.0
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0977s; samplesPerSecond = 6553.1
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.1008s; samplesPerSecond = 6348.1
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0993s; samplesPerSecond = 6443.8
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.07547s
|
||||
MPI Rank 0: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.0934s; samplesPerSecond = 6851.1
|
||||
MPI Rank 0: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0945s; samplesPerSecond = 6775.2
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.1008s; samplesPerSecond = 6351.3
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0851s; samplesPerSecond = 7522.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.1101s; samplesPerSecond = 5814.1
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0879s; samplesPerSecond = 7280.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0876s; samplesPerSecond = 7306.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.1010s; samplesPerSecond = 6338.0
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0915s; samplesPerSecond = 6997.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.1070s; samplesPerSecond = 5978.9
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0879s; samplesPerSecond = 7277.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0881s; samplesPerSecond = 7264.4
|
||||
MPI Rank 0: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0948s; samplesPerSecond = 6750.9
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0913s; samplesPerSecond = 7013.3
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0928s; samplesPerSecond = 6897.2
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.1049s; samplesPerSecond = 6100.2
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0865s; samplesPerSecond = 7398.0
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0889s; samplesPerSecond = 7202.8
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0971s; samplesPerSecond = 6594.3
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0983s; samplesPerSecond = 6511.5
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1010s; samplesPerSecond = 6339.3
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0944s; samplesPerSecond = 6782.8
|
||||
MPI Rank 0: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1063s; samplesPerSecond = 6018.7
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1045s; samplesPerSecond = 6124.6
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0902s; samplesPerSecond = 7096.6
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.1010s; samplesPerSecond = 6336.6
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0913s; samplesPerSecond = 7013.5
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0944s; samplesPerSecond = 6776.6
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0897s; samplesPerSecond = 7136.0
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0977s; samplesPerSecond = 6553.1
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.1008s; samplesPerSecond = 6348.1
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0993s; samplesPerSecond = 6443.8
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.07547s
|
||||
MPI Rank 0: 08/16/2016 03:19:31: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_Parallel1BitQuantization@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:31: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalErrorPrediction = 0.60039062 * 2560; time = 0.2084s; samplesPerSecond = 12286.1
|
||||
MPI Rank 0: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalErrorPrediction = 0.57070312 * 2560; time = 0.1804s; samplesPerSecond = 14191.5
|
||||
MPI Rank 0: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalErrorPrediction = 0.56289062 * 2560; time = 0.1704s; samplesPerSecond = 15022.6
|
||||
MPI Rank 0: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalErrorPrediction = 0.56171875 * 2560; time = 0.1680s; samplesPerSecond = 15242.1
|
||||
MPI Rank 0: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalErrorPrediction = 0.55976563 * 2560; time = 0.1697s; samplesPerSecond = 15081.4
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalErrorPrediction = 0.54531250 * 2560; time = 0.1754s; samplesPerSecond = 14597.3
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalErrorPrediction = 0.54765625 * 2560; time = 0.1695s; samplesPerSecond = 15105.3
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalErrorPrediction = 0.55507812 * 2560; time = 0.1716s; samplesPerSecond = 14922.7
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalErrorPrediction = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42511s
|
||||
MPI Rank 0: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalClassificationError = 0.60039062 * 2560; time = 0.2084s; samplesPerSecond = 12286.1
|
||||
MPI Rank 0: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalClassificationError = 0.57070312 * 2560; time = 0.1804s; samplesPerSecond = 14191.5
|
||||
MPI Rank 0: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalClassificationError = 0.56289062 * 2560; time = 0.1704s; samplesPerSecond = 15022.6
|
||||
MPI Rank 0: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalClassificationError = 0.56171875 * 2560; time = 0.1680s; samplesPerSecond = 15242.1
|
||||
MPI Rank 0: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalClassificationError = 0.55976563 * 2560; time = 0.1697s; samplesPerSecond = 15081.4
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalClassificationError = 0.54531250 * 2560; time = 0.1754s; samplesPerSecond = 14597.3
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalClassificationError = 0.54765625 * 2560; time = 0.1695s; samplesPerSecond = 15105.3
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalClassificationError = 0.55507812 * 2560; time = 0.1716s; samplesPerSecond = 14922.7
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalClassificationError = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42511s
|
||||
MPI Rank 0: 08/16/2016 03:19:33: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_Parallel1BitQuantization@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalErrorPrediction = 0.53154297 * 10240; time = 0.3564s; samplesPerSecond = 28732.2
|
||||
MPI Rank 0: 08/16/2016 03:19:34: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalErrorPrediction = 0.55019531 * 10240; time = 0.3227s; samplesPerSecond = 31728.4
|
||||
MPI Rank 0: 08/16/2016 03:19:34: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalErrorPrediction = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.689726s
|
||||
MPI Rank 0: 08/16/2016 03:19:33: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalClassificationError = 0.53154297 * 10240; time = 0.3564s; samplesPerSecond = 28732.2
|
||||
MPI Rank 0: 08/16/2016 03:19:34: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalClassificationError = 0.55019531 * 10240; time = 0.3227s; samplesPerSecond = 31728.4
|
||||
MPI Rank 0: 08/16/2016 03:19:34: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalClassificationError = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.689726s
|
||||
MPI Rank 0: 08/16/2016 03:19:34: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_Parallel1BitQuantization@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:19:34: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -687,7 +687,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -713,7 +713,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -793,7 +793,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -819,7 +819,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -905,7 +905,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -931,7 +931,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1028,7 +1028,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1057,7 +1057,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1080,7 +1080,7 @@ MPI Rank 1: 08/16/2016 03:19:23: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:19:23: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:23: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:19:23: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:19:23: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1133,61 +1133,61 @@ MPI Rank 1: 08/16/2016 03:19:28: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:28: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1401s; samplesPerSecond = 4569.0
|
||||
MPI Rank 1: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0940s; samplesPerSecond = 6807.0
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.1207s; samplesPerSecond = 5302.3
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0910s; samplesPerSecond = 7036.1
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.1037s; samplesPerSecond = 6172.4
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0947s; samplesPerSecond = 6758.8
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0943s; samplesPerSecond = 6788.4
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.1106s; samplesPerSecond = 5784.9
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0876s; samplesPerSecond = 7307.6
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.1082s; samplesPerSecond = 5917.2
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.1046s; samplesPerSecond = 6117.8
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.1133s; samplesPerSecond = 5647.6
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0890s; samplesPerSecond = 7191.7
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.1012s; samplesPerSecond = 6323.4
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.1044s; samplesPerSecond = 6132.4
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.1080s; samplesPerSecond = 5928.1
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.1107s; samplesPerSecond = 5782.9
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1113s; samplesPerSecond = 5752.3
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.1110s; samplesPerSecond = 5764.9
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.1045s; samplesPerSecond = 6122.8
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0979s; samplesPerSecond = 6535.5
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1039s; samplesPerSecond = 6161.4
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0918s; samplesPerSecond = 6972.7
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0950s; samplesPerSecond = 6734.1
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.1074s; samplesPerSecond = 5959.5
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0968s; samplesPerSecond = 6610.5
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1041s; samplesPerSecond = 6148.1
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.1035s; samplesPerSecond = 6185.8
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0917s; samplesPerSecond = 6977.2
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0773s; samplesPerSecond = 8284.0
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0373s; samplesPerSecond = 17172.9
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0373s; samplesPerSecond = 17176.1
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16555s
|
||||
MPI Rank 1: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1401s; samplesPerSecond = 4569.0
|
||||
MPI Rank 1: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0940s; samplesPerSecond = 6807.0
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.1207s; samplesPerSecond = 5302.3
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0910s; samplesPerSecond = 7036.1
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.1037s; samplesPerSecond = 6172.4
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0947s; samplesPerSecond = 6758.8
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0943s; samplesPerSecond = 6788.4
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.1106s; samplesPerSecond = 5784.9
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0876s; samplesPerSecond = 7307.6
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.1082s; samplesPerSecond = 5917.2
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.1046s; samplesPerSecond = 6117.8
|
||||
MPI Rank 1: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.1133s; samplesPerSecond = 5647.6
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0890s; samplesPerSecond = 7191.7
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.1012s; samplesPerSecond = 6323.4
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.1044s; samplesPerSecond = 6132.4
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.1080s; samplesPerSecond = 5928.1
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.1107s; samplesPerSecond = 5782.9
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1113s; samplesPerSecond = 5752.3
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.1110s; samplesPerSecond = 5764.9
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.1045s; samplesPerSecond = 6122.8
|
||||
MPI Rank 1: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0979s; samplesPerSecond = 6535.5
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1039s; samplesPerSecond = 6161.4
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0918s; samplesPerSecond = 6972.7
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0950s; samplesPerSecond = 6734.1
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.1074s; samplesPerSecond = 5959.5
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0968s; samplesPerSecond = 6610.5
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1041s; samplesPerSecond = 6148.1
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.1035s; samplesPerSecond = 6185.8
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0917s; samplesPerSecond = 6977.2
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0773s; samplesPerSecond = 8284.0
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0373s; samplesPerSecond = 17172.9
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0373s; samplesPerSecond = 17176.1
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16555s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:31: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalErrorPrediction = 0.60039062 * 2560; time = 0.2063s; samplesPerSecond = 12411.0
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalErrorPrediction = 0.57070312 * 2560; time = 0.1808s; samplesPerSecond = 14158.0
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalErrorPrediction = 0.56289062 * 2560; time = 0.1701s; samplesPerSecond = 15049.2
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalErrorPrediction = 0.56171875 * 2560; time = 0.1683s; samplesPerSecond = 15215.0
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalErrorPrediction = 0.55976563 * 2560; time = 0.1697s; samplesPerSecond = 15086.2
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalErrorPrediction = 0.54531250 * 2560; time = 0.1754s; samplesPerSecond = 14598.1
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalErrorPrediction = 0.54765625 * 2560; time = 0.1692s; samplesPerSecond = 15131.5
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalErrorPrediction = 0.55507812 * 2560; time = 0.1718s; samplesPerSecond = 14902.8
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalErrorPrediction = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42565s
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalClassificationError = 0.60039062 * 2560; time = 0.2063s; samplesPerSecond = 12411.0
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalClassificationError = 0.57070312 * 2560; time = 0.1808s; samplesPerSecond = 14158.0
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalClassificationError = 0.56289062 * 2560; time = 0.1701s; samplesPerSecond = 15049.2
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalClassificationError = 0.56171875 * 2560; time = 0.1683s; samplesPerSecond = 15215.0
|
||||
MPI Rank 1: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalClassificationError = 0.55976563 * 2560; time = 0.1697s; samplesPerSecond = 15086.2
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalClassificationError = 0.54531250 * 2560; time = 0.1754s; samplesPerSecond = 14598.1
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalClassificationError = 0.54765625 * 2560; time = 0.1692s; samplesPerSecond = 15131.5
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalClassificationError = 0.55507812 * 2560; time = 0.1718s; samplesPerSecond = 14902.8
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalClassificationError = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42565s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalErrorPrediction = 0.53154297 * 10240; time = 0.3534s; samplesPerSecond = 28972.3
|
||||
MPI Rank 1: 08/16/2016 03:19:34: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalErrorPrediction = 0.55019531 * 10240; time = 0.3227s; samplesPerSecond = 31731.1
|
||||
MPI Rank 1: 08/16/2016 03:19:34: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalErrorPrediction = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.690282s
|
||||
MPI Rank 1: 08/16/2016 03:19:33: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalClassificationError = 0.53154297 * 10240; time = 0.3534s; samplesPerSecond = 28972.3
|
||||
MPI Rank 1: 08/16/2016 03:19:34: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalClassificationError = 0.55019531 * 10240; time = 0.3227s; samplesPerSecond = 31731.1
|
||||
MPI Rank 1: 08/16/2016 03:19:34: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalClassificationError = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.690282s
|
||||
MPI Rank 1: 08/16/2016 03:19:34: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:34: Action "train" complete.
|
||||
|
|
@ -1240,7 +1240,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1266,7 +1266,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1346,7 +1346,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1372,7 +1372,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1458,7 +1458,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1484,7 +1484,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1581,7 +1581,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1610,7 +1610,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1633,7 +1633,7 @@ MPI Rank 2: 08/16/2016 03:19:23: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 03:19:23: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:23: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 03:19:23: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 03:19:23: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1686,61 +1686,61 @@ MPI Rank 2: 08/16/2016 03:19:28: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:28: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.0912s; samplesPerSecond = 7021.2
|
||||
MPI Rank 2: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0956s; samplesPerSecond = 6692.3
|
||||
MPI Rank 2: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0836s; samplesPerSecond = 7652.6
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0959s; samplesPerSecond = 6671.7
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0910s; samplesPerSecond = 7030.3
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.1035s; samplesPerSecond = 6184.1
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.1052s; samplesPerSecond = 6083.4
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.0975s; samplesPerSecond = 6564.4
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0947s; samplesPerSecond = 6758.0
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0973s; samplesPerSecond = 6577.9
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0949s; samplesPerSecond = 6744.6
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.1008s; samplesPerSecond = 6346.3
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0908s; samplesPerSecond = 7044.7
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.1112s; samplesPerSecond = 5753.2
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0946s; samplesPerSecond = 6763.8
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0886s; samplesPerSecond = 7225.6
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.1048s; samplesPerSecond = 6108.6
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0928s; samplesPerSecond = 6895.7
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0930s; samplesPerSecond = 6882.8
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0850s; samplesPerSecond = 7533.0
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0877s; samplesPerSecond = 7301.7
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0880s; samplesPerSecond = 7269.5
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0909s; samplesPerSecond = 7038.5
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0855s; samplesPerSecond = 7482.3
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.1141s; samplesPerSecond = 5608.1
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0974s; samplesPerSecond = 6573.7
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0854s; samplesPerSecond = 7498.0
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.1040s; samplesPerSecond = 6154.6
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.1073s; samplesPerSecond = 5966.9
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0921s; samplesPerSecond = 6948.4
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0959s; samplesPerSecond = 6674.3
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0907s; samplesPerSecond = 7055.5
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.06702s
|
||||
MPI Rank 2: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.0912s; samplesPerSecond = 7021.2
|
||||
MPI Rank 2: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0956s; samplesPerSecond = 6692.3
|
||||
MPI Rank 2: 08/16/2016 03:19:28: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0836s; samplesPerSecond = 7652.6
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0959s; samplesPerSecond = 6671.7
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0910s; samplesPerSecond = 7030.3
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.1035s; samplesPerSecond = 6184.1
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.1052s; samplesPerSecond = 6083.4
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.0975s; samplesPerSecond = 6564.4
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0947s; samplesPerSecond = 6758.0
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0973s; samplesPerSecond = 6577.9
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0949s; samplesPerSecond = 6744.6
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.1008s; samplesPerSecond = 6346.3
|
||||
MPI Rank 2: 08/16/2016 03:19:29: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0908s; samplesPerSecond = 7044.7
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.1112s; samplesPerSecond = 5753.2
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0946s; samplesPerSecond = 6763.8
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0886s; samplesPerSecond = 7225.6
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.1048s; samplesPerSecond = 6108.6
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0928s; samplesPerSecond = 6895.7
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0930s; samplesPerSecond = 6882.8
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0850s; samplesPerSecond = 7533.0
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0877s; samplesPerSecond = 7301.7
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0880s; samplesPerSecond = 7269.5
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0909s; samplesPerSecond = 7038.5
|
||||
MPI Rank 2: 08/16/2016 03:19:30: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0855s; samplesPerSecond = 7482.3
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.1141s; samplesPerSecond = 5608.1
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0974s; samplesPerSecond = 6573.7
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0854s; samplesPerSecond = 7498.0
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.1040s; samplesPerSecond = 6154.6
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.1073s; samplesPerSecond = 5966.9
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0921s; samplesPerSecond = 6948.4
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0959s; samplesPerSecond = 6674.3
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0907s; samplesPerSecond = 7055.5
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.06702s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:31: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalErrorPrediction = 0.60039062 * 2560; time = 0.2089s; samplesPerSecond = 12254.0
|
||||
MPI Rank 2: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalErrorPrediction = 0.57070312 * 2560; time = 0.1802s; samplesPerSecond = 14210.1
|
||||
MPI Rank 2: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalErrorPrediction = 0.56289062 * 2560; time = 0.1701s; samplesPerSecond = 15046.4
|
||||
MPI Rank 2: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalErrorPrediction = 0.56171875 * 2560; time = 0.1684s; samplesPerSecond = 15206.1
|
||||
MPI Rank 2: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalErrorPrediction = 0.55976563 * 2560; time = 0.1697s; samplesPerSecond = 15085.8
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalErrorPrediction = 0.54531250 * 2560; time = 0.1754s; samplesPerSecond = 14595.8
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalErrorPrediction = 0.54765625 * 2560; time = 0.1692s; samplesPerSecond = 15129.6
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalErrorPrediction = 0.55507812 * 2560; time = 0.1718s; samplesPerSecond = 14899.1
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalErrorPrediction = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42528s
|
||||
MPI Rank 2: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.19429672 * 2560; EvalClassificationError = 0.60039062 * 2560; time = 0.2089s; samplesPerSecond = 12254.0
|
||||
MPI Rank 2: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.15577544 * 2560; EvalClassificationError = 0.57070312 * 2560; time = 0.1802s; samplesPerSecond = 14210.1
|
||||
MPI Rank 2: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.09655267 * 2560; EvalClassificationError = 0.56289062 * 2560; time = 0.1701s; samplesPerSecond = 15046.4
|
||||
MPI Rank 2: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.06745115 * 2560; EvalClassificationError = 0.56171875 * 2560; time = 0.1684s; samplesPerSecond = 15206.1
|
||||
MPI Rank 2: 08/16/2016 03:19:32: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.06705242 * 2560; EvalClassificationError = 0.55976563 * 2560; time = 0.1697s; samplesPerSecond = 15085.8
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.00136482 * 2560; EvalClassificationError = 0.54531250 * 2560; time = 0.1754s; samplesPerSecond = 14595.8
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99508116 * 2560; EvalClassificationError = 0.54765625 * 2560; time = 0.1692s; samplesPerSecond = 15129.6
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.99964996 * 2560; EvalClassificationError = 0.55507812 * 2560; time = 0.1718s; samplesPerSecond = 14899.1
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.07215304 * 20480; EvalClassificationError = 0.56293945 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42528s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 1), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalErrorPrediction = 0.53154297 * 10240; time = 0.3569s; samplesPerSecond = 28689.5
|
||||
MPI Rank 2: 08/16/2016 03:19:34: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalErrorPrediction = 0.55019531 * 10240; time = 0.3228s; samplesPerSecond = 31727.1
|
||||
MPI Rank 2: 08/16/2016 03:19:34: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalErrorPrediction = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.689913s
|
||||
MPI Rank 2: 08/16/2016 03:19:33: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95876979 * 10240; EvalClassificationError = 0.53154297 * 10240; time = 0.3569s; samplesPerSecond = 28689.5
|
||||
MPI Rank 2: 08/16/2016 03:19:34: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97868556 * 10240; EvalClassificationError = 0.55019531 * 10240; time = 0.3228s; samplesPerSecond = 31727.1
|
||||
MPI Rank 2: 08/16/2016 03:19:34: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.96872768 * 20480; EvalClassificationError = 0.54086914 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.689913s
|
||||
MPI Rank 2: 08/16/2016 03:19:34: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:19:34: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -105,7 +105,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -131,7 +131,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -213,7 +213,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -239,7 +239,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -327,7 +327,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -353,7 +353,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -452,7 +452,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -481,7 +481,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -504,7 +504,7 @@ MPI Rank 0: 08/16/2016 10:08:13: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 10:08:13: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:13: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 10:08:13: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 10:08:13: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -557,113 +557,113 @@ MPI Rank 0: 08/16/2016 10:08:15: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:15: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.62930672 * 192; EvalErrorPrediction = 0.90625000 * 192; time = 0.1034s; samplesPerSecond = 1856.4
|
||||
MPI Rank 0: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.39189508 * 192; EvalErrorPrediction = 0.88020833 * 192; time = 0.0957s; samplesPerSecond = 2005.3
|
||||
MPI Rank 0: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.73456191 * 192; EvalErrorPrediction = 0.92708333 * 192; time = 0.0961s; samplesPerSecond = 1998.5
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.22329144 * 192; EvalErrorPrediction = 0.91145833 * 192; time = 0.0976s; samplesPerSecond = 1966.2
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.44791490 * 192; EvalErrorPrediction = 0.90625000 * 192; time = 0.0961s; samplesPerSecond = 1998.4
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.62%]: CrossEntropyWithSoftmax = 4.37559721 * 192; EvalErrorPrediction = 0.93750000 * 192; time = 0.0993s; samplesPerSecond = 1933.4
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.24117585 * 192; EvalErrorPrediction = 0.94791667 * 192; time = 0.0996s; samplesPerSecond = 1927.7
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.05463916 * 192; EvalErrorPrediction = 0.89062500 * 192; time = 0.1003s; samplesPerSecond = 1915.1
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.98830106 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0989s; samplesPerSecond = 1942.3
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.84560829 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0982s; samplesPerSecond = 1954.9
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.81542483 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0935s; samplesPerSecond = 2054.4
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.75181899 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0960s; samplesPerSecond = 2000.2
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.74962983 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0985s; samplesPerSecond = 1949.4
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.12%]: CrossEntropyWithSoftmax = 3.70309729 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0970s; samplesPerSecond = 1978.9
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89401044 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0969s; samplesPerSecond = 1981.3
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.78605845 * 192; EvalErrorPrediction = 0.83854167 * 192; time = 0.1005s; samplesPerSecond = 1909.6
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.82243497 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.1019s; samplesPerSecond = 1883.3
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.91866810 * 192; EvalErrorPrediction = 0.85937500 * 192; time = 0.0931s; samplesPerSecond = 2061.2
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56637464 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.1001s; samplesPerSecond = 1917.6
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.64635888 * 192; EvalErrorPrediction = 0.86458333 * 192; time = 0.0994s; samplesPerSecond = 1931.2
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.35602557 * 192; EvalErrorPrediction = 0.75520833 * 192; time = 0.0961s; samplesPerSecond = 1997.1
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.62%]: CrossEntropyWithSoftmax = 3.40627666 * 192; EvalErrorPrediction = 0.78645833 * 192; time = 0.0923s; samplesPerSecond = 2079.6
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.35477965 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0966s; samplesPerSecond = 1986.8
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.66184880 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0993s; samplesPerSecond = 1934.5
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.62630442 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0984s; samplesPerSecond = 1951.2
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.36376118 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0993s; samplesPerSecond = 1933.0
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51072509 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0990s; samplesPerSecond = 1938.9
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47385790 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.1012s; samplesPerSecond = 1897.7
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.55227399 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0987s; samplesPerSecond = 1946.3
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.12%]: CrossEntropyWithSoftmax = 3.44007374 * 192; EvalErrorPrediction = 0.81250000 * 192; time = 0.0984s; samplesPerSecond = 1951.8
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.35761615 * 192; EvalErrorPrediction = 0.78645833 * 192; time = 0.0960s; samplesPerSecond = 2000.5
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55772259 * 192; EvalErrorPrediction = 0.81250000 * 192; time = 0.0959s; samplesPerSecond = 2002.8
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.33538937 * 192; EvalErrorPrediction = 0.81250000 * 192; time = 0.0974s; samplesPerSecond = 1971.4
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.51412750 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.1002s; samplesPerSecond = 1916.1
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.56303150 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0981s; samplesPerSecond = 1957.0
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.48659009 * 192; EvalErrorPrediction = 0.83854167 * 192; time = 0.0991s; samplesPerSecond = 1937.2
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.33311671 * 192; EvalErrorPrediction = 0.80208333 * 192; time = 0.0998s; samplesPerSecond = 1924.6
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.62%]: CrossEntropyWithSoftmax = 3.28125589 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0954s; samplesPerSecond = 2012.5
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.20377310 * 192; EvalErrorPrediction = 0.78645833 * 192; time = 0.0995s; samplesPerSecond = 1930.2
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27073629 * 192; EvalErrorPrediction = 0.78125000 * 192; time = 0.0990s; samplesPerSecond = 1940.3
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.55050536 * 192; EvalErrorPrediction = 0.83854167 * 192; time = 0.0981s; samplesPerSecond = 1956.9
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00419998 * 192; EvalErrorPrediction = 0.71354167 * 192; time = 0.0981s; samplesPerSecond = 1956.3
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.11647605 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.1036s; samplesPerSecond = 1852.6
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.06195640 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0961s; samplesPerSecond = 1997.5
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18915458 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0973s; samplesPerSecond = 1973.9
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.12%]: CrossEntropyWithSoftmax = 2.77654930 * 192; EvalErrorPrediction = 0.71354167 * 192; time = 0.0989s; samplesPerSecond = 1942.0
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17408570 * 192; EvalErrorPrediction = 0.83854167 * 192; time = 0.1002s; samplesPerSecond = 1915.9
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03022229 * 192; EvalErrorPrediction = 0.72916667 * 192; time = 0.0957s; samplesPerSecond = 2005.6
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.01692112 * 192; EvalErrorPrediction = 0.74479167 * 192; time = 0.0964s; samplesPerSecond = 1991.1
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79519565 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0961s; samplesPerSecond = 1997.2
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.03721970 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0982s; samplesPerSecond = 1954.6
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.16911303 * 192; EvalErrorPrediction = 0.78645833 * 192; time = 0.0957s; samplesPerSecond = 2006.0
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11747666 * 192; EvalErrorPrediction = 0.75000000 * 192; time = 0.1003s; samplesPerSecond = 1914.3
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.62%]: CrossEntropyWithSoftmax = 2.86999165 * 192; EvalErrorPrediction = 0.72395833 * 192; time = 0.0998s; samplesPerSecond = 1923.8
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.78866812 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0975s; samplesPerSecond = 1968.3
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00684431 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.1030s; samplesPerSecond = 1863.8
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.72612865 * 192; EvalErrorPrediction = 0.68229167 * 192; time = 0.0995s; samplesPerSecond = 1929.9
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64981655 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0999s; samplesPerSecond = 1921.9
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71554609 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0996s; samplesPerSecond = 1926.9
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.73031896 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0979s; samplesPerSecond = 1960.9
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.83317991 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0996s; samplesPerSecond = 1927.5
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79636434 * 192; EvalErrorPrediction = 0.71354167 * 192; time = 0.0991s; samplesPerSecond = 1938.1
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.77040456 * 192; EvalErrorPrediction = 0.72395833 * 192; time = 0.0994s; samplesPerSecond = 1932.0
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.79003220 * 192; EvalErrorPrediction = 0.69791667 * 192; time = 0.0958s; samplesPerSecond = 2003.2
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.54457731 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0980s; samplesPerSecond = 1960.1
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97922473 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0958s; samplesPerSecond = 2004.4
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.62424232 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0975s; samplesPerSecond = 1970.2
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52182003 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0949s; samplesPerSecond = 2024.0
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.60628500 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0960s; samplesPerSecond = 1999.8
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.62%]: CrossEntropyWithSoftmax = 2.57508979 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0995s; samplesPerSecond = 1929.9
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.57480702 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0989s; samplesPerSecond = 1941.7
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.47514723 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0968s; samplesPerSecond = 1982.8
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.73668923 * 192; EvalErrorPrediction = 0.67187500 * 192; time = 0.0966s; samplesPerSecond = 1987.1
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.47244978 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0977s; samplesPerSecond = 1964.8
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.43386779 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.1001s; samplesPerSecond = 1917.8
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.71554733 * 192; EvalErrorPrediction = 0.68750000 * 192; time = 0.0983s; samplesPerSecond = 1953.8
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.54334345 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0997s; samplesPerSecond = 1924.9
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.12%]: CrossEntropyWithSoftmax = 2.97324196 * 192; EvalErrorPrediction = 0.76041667 * 192; time = 0.1000s; samplesPerSecond = 1920.7
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.33516568 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0980s; samplesPerSecond = 1959.3
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.23213800 * 192; EvalErrorPrediction = 0.60937500 * 192; time = 0.0952s; samplesPerSecond = 2017.0
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.29168780 * 192; EvalErrorPrediction = 0.57812500 * 192; time = 0.0935s; samplesPerSecond = 2053.8
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.68110845 * 192; EvalErrorPrediction = 0.70312500 * 192; time = 0.0949s; samplesPerSecond = 2023.3
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.43379863 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0934s; samplesPerSecond = 2055.5
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.48936113 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0938s; samplesPerSecond = 2046.7
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.37855205 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0956s; samplesPerSecond = 2009.0
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.62%]: CrossEntropyWithSoftmax = 2.35951804 * 192; EvalErrorPrediction = 0.57812500 * 192; time = 0.0991s; samplesPerSecond = 1937.2
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.32828318 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0964s; samplesPerSecond = 1992.0
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.27181872 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0971s; samplesPerSecond = 1976.8
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.07018832 * 192; EvalErrorPrediction = 0.50520833 * 192; time = 0.0960s; samplesPerSecond = 2000.8
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.40647801 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0970s; samplesPerSecond = 1979.9
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.51858879 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0928s; samplesPerSecond = 2067.9
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33656120 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0974s; samplesPerSecond = 1971.4
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.60987825 * 192; EvalErrorPrediction = 0.69791667 * 192; time = 0.0983s; samplesPerSecond = 1952.9
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.12%]: CrossEntropyWithSoftmax = 2.46713772 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.1023s; samplesPerSecond = 1877.2
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30355405 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.1012s; samplesPerSecond = 1896.9
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.19148527 * 192; EvalErrorPrediction = 0.59895833 * 192; time = 0.1033s; samplesPerSecond = 1858.4
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.25368531 * 192; EvalErrorPrediction = 0.56250000 * 192; time = 0.0975s; samplesPerSecond = 1969.8
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.10763074 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.1009s; samplesPerSecond = 1902.3
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.30064187 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.1004s; samplesPerSecond = 1912.9
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.27695669 * 192; EvalErrorPrediction = 0.59375000 * 192; time = 0.1013s; samplesPerSecond = 1895.8
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.13317518 * 192; EvalErrorPrediction = 0.56250000 * 192; time = 0.1026s; samplesPerSecond = 1872.0
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.62%]: CrossEntropyWithSoftmax = 2.23952900 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0997s; samplesPerSecond = 1926.4
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.13404008 * 192; EvalErrorPrediction = 0.54687500 * 192; time = 0.0992s; samplesPerSecond = 1935.5
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.42845170 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0962s; samplesPerSecond = 1995.2
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.15843921 * 192; EvalErrorPrediction = 0.60937500 * 192; time = 0.0957s; samplesPerSecond = 2005.7
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.20218732 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0968s; samplesPerSecond = 1983.0
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=10.4744s
|
||||
MPI Rank 0: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.62930672 * 192; EvalClassificationError = 0.90625000 * 192; time = 0.1034s; samplesPerSecond = 1856.4
|
||||
MPI Rank 0: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.39189508 * 192; EvalClassificationError = 0.88020833 * 192; time = 0.0957s; samplesPerSecond = 2005.3
|
||||
MPI Rank 0: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.73456191 * 192; EvalClassificationError = 0.92708333 * 192; time = 0.0961s; samplesPerSecond = 1998.5
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.22329144 * 192; EvalClassificationError = 0.91145833 * 192; time = 0.0976s; samplesPerSecond = 1966.2
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.44791490 * 192; EvalClassificationError = 0.90625000 * 192; time = 0.0961s; samplesPerSecond = 1998.4
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.62%]: CrossEntropyWithSoftmax = 4.37559721 * 192; EvalClassificationError = 0.93750000 * 192; time = 0.0993s; samplesPerSecond = 1933.4
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.24117585 * 192; EvalClassificationError = 0.94791667 * 192; time = 0.0996s; samplesPerSecond = 1927.7
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.05463916 * 192; EvalClassificationError = 0.89062500 * 192; time = 0.1003s; samplesPerSecond = 1915.1
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.98830106 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0989s; samplesPerSecond = 1942.3
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.84560829 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0982s; samplesPerSecond = 1954.9
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.81542483 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0935s; samplesPerSecond = 2054.4
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.75181899 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0960s; samplesPerSecond = 2000.2
|
||||
MPI Rank 0: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.74962983 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0985s; samplesPerSecond = 1949.4
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.12%]: CrossEntropyWithSoftmax = 3.70309729 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0970s; samplesPerSecond = 1978.9
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89401044 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0969s; samplesPerSecond = 1981.3
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.78605845 * 192; EvalClassificationError = 0.83854167 * 192; time = 0.1005s; samplesPerSecond = 1909.6
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.82243497 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.1019s; samplesPerSecond = 1883.3
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.91866810 * 192; EvalClassificationError = 0.85937500 * 192; time = 0.0931s; samplesPerSecond = 2061.2
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56637464 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.1001s; samplesPerSecond = 1917.6
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.64635888 * 192; EvalClassificationError = 0.86458333 * 192; time = 0.0994s; samplesPerSecond = 1931.2
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.35602557 * 192; EvalClassificationError = 0.75520833 * 192; time = 0.0961s; samplesPerSecond = 1997.1
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.62%]: CrossEntropyWithSoftmax = 3.40627666 * 192; EvalClassificationError = 0.78645833 * 192; time = 0.0923s; samplesPerSecond = 2079.6
|
||||
MPI Rank 0: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.35477965 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0966s; samplesPerSecond = 1986.8
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.66184880 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0993s; samplesPerSecond = 1934.5
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.62630442 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0984s; samplesPerSecond = 1951.2
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.36376118 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0993s; samplesPerSecond = 1933.0
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51072509 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0990s; samplesPerSecond = 1938.9
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47385790 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.1012s; samplesPerSecond = 1897.7
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.55227399 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0987s; samplesPerSecond = 1946.3
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.12%]: CrossEntropyWithSoftmax = 3.44007374 * 192; EvalClassificationError = 0.81250000 * 192; time = 0.0984s; samplesPerSecond = 1951.8
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.35761615 * 192; EvalClassificationError = 0.78645833 * 192; time = 0.0960s; samplesPerSecond = 2000.5
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55772259 * 192; EvalClassificationError = 0.81250000 * 192; time = 0.0959s; samplesPerSecond = 2002.8
|
||||
MPI Rank 0: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.33538937 * 192; EvalClassificationError = 0.81250000 * 192; time = 0.0974s; samplesPerSecond = 1971.4
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.51412750 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.1002s; samplesPerSecond = 1916.1
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.56303150 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0981s; samplesPerSecond = 1957.0
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.48659009 * 192; EvalClassificationError = 0.83854167 * 192; time = 0.0991s; samplesPerSecond = 1937.2
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.33311671 * 192; EvalClassificationError = 0.80208333 * 192; time = 0.0998s; samplesPerSecond = 1924.6
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.62%]: CrossEntropyWithSoftmax = 3.28125589 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0954s; samplesPerSecond = 2012.5
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.20377310 * 192; EvalClassificationError = 0.78645833 * 192; time = 0.0995s; samplesPerSecond = 1930.2
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27073629 * 192; EvalClassificationError = 0.78125000 * 192; time = 0.0990s; samplesPerSecond = 1940.3
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.55050536 * 192; EvalClassificationError = 0.83854167 * 192; time = 0.0981s; samplesPerSecond = 1956.9
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00419998 * 192; EvalClassificationError = 0.71354167 * 192; time = 0.0981s; samplesPerSecond = 1956.3
|
||||
MPI Rank 0: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.11647605 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.1036s; samplesPerSecond = 1852.6
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.06195640 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0961s; samplesPerSecond = 1997.5
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18915458 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0973s; samplesPerSecond = 1973.9
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.12%]: CrossEntropyWithSoftmax = 2.77654930 * 192; EvalClassificationError = 0.71354167 * 192; time = 0.0989s; samplesPerSecond = 1942.0
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17408570 * 192; EvalClassificationError = 0.83854167 * 192; time = 0.1002s; samplesPerSecond = 1915.9
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03022229 * 192; EvalClassificationError = 0.72916667 * 192; time = 0.0957s; samplesPerSecond = 2005.6
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.01692112 * 192; EvalClassificationError = 0.74479167 * 192; time = 0.0964s; samplesPerSecond = 1991.1
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79519565 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0961s; samplesPerSecond = 1997.2
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.03721970 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0982s; samplesPerSecond = 1954.6
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.16911303 * 192; EvalClassificationError = 0.78645833 * 192; time = 0.0957s; samplesPerSecond = 2006.0
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11747666 * 192; EvalClassificationError = 0.75000000 * 192; time = 0.1003s; samplesPerSecond = 1914.3
|
||||
MPI Rank 0: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.62%]: CrossEntropyWithSoftmax = 2.86999165 * 192; EvalClassificationError = 0.72395833 * 192; time = 0.0998s; samplesPerSecond = 1923.8
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.78866812 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0975s; samplesPerSecond = 1968.3
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00684431 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.1030s; samplesPerSecond = 1863.8
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.72612865 * 192; EvalClassificationError = 0.68229167 * 192; time = 0.0995s; samplesPerSecond = 1929.9
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64981655 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0999s; samplesPerSecond = 1921.9
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71554609 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0996s; samplesPerSecond = 1926.9
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.73031896 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0979s; samplesPerSecond = 1960.9
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.83317991 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0996s; samplesPerSecond = 1927.5
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79636434 * 192; EvalClassificationError = 0.71354167 * 192; time = 0.0991s; samplesPerSecond = 1938.1
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.77040456 * 192; EvalClassificationError = 0.72395833 * 192; time = 0.0994s; samplesPerSecond = 1932.0
|
||||
MPI Rank 0: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.79003220 * 192; EvalClassificationError = 0.69791667 * 192; time = 0.0958s; samplesPerSecond = 2003.2
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.54457731 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0980s; samplesPerSecond = 1960.1
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97922473 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0958s; samplesPerSecond = 2004.4
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.62424232 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0975s; samplesPerSecond = 1970.2
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52182003 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0949s; samplesPerSecond = 2024.0
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.60628500 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0960s; samplesPerSecond = 1999.8
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.62%]: CrossEntropyWithSoftmax = 2.57508979 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0995s; samplesPerSecond = 1929.9
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.57480702 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0989s; samplesPerSecond = 1941.7
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.47514723 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0968s; samplesPerSecond = 1982.8
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.73668923 * 192; EvalClassificationError = 0.67187500 * 192; time = 0.0966s; samplesPerSecond = 1987.1
|
||||
MPI Rank 0: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.47244978 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0977s; samplesPerSecond = 1964.8
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.43386779 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.1001s; samplesPerSecond = 1917.8
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.71554733 * 192; EvalClassificationError = 0.68750000 * 192; time = 0.0983s; samplesPerSecond = 1953.8
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.54334345 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0997s; samplesPerSecond = 1924.9
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.12%]: CrossEntropyWithSoftmax = 2.97324196 * 192; EvalClassificationError = 0.76041667 * 192; time = 0.1000s; samplesPerSecond = 1920.7
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.33516568 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0980s; samplesPerSecond = 1959.3
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.23213800 * 192; EvalClassificationError = 0.60937500 * 192; time = 0.0952s; samplesPerSecond = 2017.0
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.29168780 * 192; EvalClassificationError = 0.57812500 * 192; time = 0.0935s; samplesPerSecond = 2053.8
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.68110845 * 192; EvalClassificationError = 0.70312500 * 192; time = 0.0949s; samplesPerSecond = 2023.3
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.43379863 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0934s; samplesPerSecond = 2055.5
|
||||
MPI Rank 0: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.48936113 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0938s; samplesPerSecond = 2046.7
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.37855205 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0956s; samplesPerSecond = 2009.0
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.62%]: CrossEntropyWithSoftmax = 2.35951804 * 192; EvalClassificationError = 0.57812500 * 192; time = 0.0991s; samplesPerSecond = 1937.2
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.32828318 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0964s; samplesPerSecond = 1992.0
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.27181872 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0971s; samplesPerSecond = 1976.8
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.07018832 * 192; EvalClassificationError = 0.50520833 * 192; time = 0.0960s; samplesPerSecond = 2000.8
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.40647801 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0970s; samplesPerSecond = 1979.9
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.51858879 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0928s; samplesPerSecond = 2067.9
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33656120 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0974s; samplesPerSecond = 1971.4
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.60987825 * 192; EvalClassificationError = 0.69791667 * 192; time = 0.0983s; samplesPerSecond = 1952.9
|
||||
MPI Rank 0: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.12%]: CrossEntropyWithSoftmax = 2.46713772 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.1023s; samplesPerSecond = 1877.2
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30355405 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.1012s; samplesPerSecond = 1896.9
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.19148527 * 192; EvalClassificationError = 0.59895833 * 192; time = 0.1033s; samplesPerSecond = 1858.4
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.25368531 * 192; EvalClassificationError = 0.56250000 * 192; time = 0.0975s; samplesPerSecond = 1969.8
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.10763074 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.1009s; samplesPerSecond = 1902.3
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.30064187 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.1004s; samplesPerSecond = 1912.9
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.27695669 * 192; EvalClassificationError = 0.59375000 * 192; time = 0.1013s; samplesPerSecond = 1895.8
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.13317518 * 192; EvalClassificationError = 0.56250000 * 192; time = 0.1026s; samplesPerSecond = 1872.0
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.62%]: CrossEntropyWithSoftmax = 2.23952900 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0997s; samplesPerSecond = 1926.4
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.13404008 * 192; EvalClassificationError = 0.54687500 * 192; time = 0.0992s; samplesPerSecond = 1935.5
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.42845170 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0962s; samplesPerSecond = 1995.2
|
||||
MPI Rank 0: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.15843921 * 192; EvalClassificationError = 0.60937500 * 192; time = 0.0957s; samplesPerSecond = 2005.7
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.20218732 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0968s; samplesPerSecond = 1983.0
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=10.4744s
|
||||
MPI Rank 0: 08/16/2016 10:08:26: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBM@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -671,46 +671,46 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.18963185 * 508; EvalErrorPrediction = 0.59842520 * 508; time = 0.2465s; samplesPerSecond = 2060.6
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.18296594 * 492; EvalErrorPrediction = 0.59146341 * 492; time = 0.2232s; samplesPerSecond = 2204.1
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.16178609 * 488; EvalErrorPrediction = 0.60040984 * 488; time = 0.2300s; samplesPerSecond = 2121.5
|
||||
MPI Rank 0: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.21470678 * 527; EvalErrorPrediction = 0.58823529 * 527; time = 0.2469s; samplesPerSecond = 2134.5
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.18963185 * 508; EvalClassificationError = 0.59842520 * 508; time = 0.2465s; samplesPerSecond = 2060.6
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.18296594 * 492; EvalClassificationError = 0.59146341 * 492; time = 0.2232s; samplesPerSecond = 2204.1
|
||||
MPI Rank 0: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.16178609 * 488; EvalClassificationError = 0.60040984 * 488; time = 0.2300s; samplesPerSecond = 2121.5
|
||||
MPI Rank 0: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.21470678 * 527; EvalClassificationError = 0.58823529 * 527; time = 0.2469s; samplesPerSecond = 2134.5
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.05-seconds latency this time; accumulated time on sync point = 0.05 seconds , average latency = 0.05 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 1.09 seconds since last report (0.00 seconds on comm.); 4289 samples processed by 2 workers (2163 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 3.94k samplesPerSecond , throughputPerWorker = 1.97k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 1.99163378 * 473; EvalErrorPrediction = 0.53276956 * 473; time = 0.2937s; samplesPerSecond = 1610.5
|
||||
MPI Rank 0: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.07069146 * 511; EvalErrorPrediction = 0.54990215 * 511; time = 0.2324s; samplesPerSecond = 2198.9
|
||||
MPI Rank 0: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.06864910 * 506; EvalErrorPrediction = 0.55138340 * 506; time = 0.2341s; samplesPerSecond = 2161.7
|
||||
MPI Rank 0: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.10498362 * 513; EvalErrorPrediction = 0.54580897 * 513; time = 0.2388s; samplesPerSecond = 2148.0
|
||||
MPI Rank 0: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 1.99163378 * 473; EvalClassificationError = 0.53276956 * 473; time = 0.2937s; samplesPerSecond = 1610.5
|
||||
MPI Rank 0: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.07069146 * 511; EvalClassificationError = 0.54990215 * 511; time = 0.2324s; samplesPerSecond = 2198.9
|
||||
MPI Rank 0: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.06864910 * 506; EvalClassificationError = 0.55138340 * 506; time = 0.2341s; samplesPerSecond = 2161.7
|
||||
MPI Rank 0: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.10498362 * 513; EvalClassificationError = 0.54580897 * 513; time = 0.2388s; samplesPerSecond = 2148.0
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.05 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 1.03 seconds since last report (0.00 seconds on comm.); 4253 samples processed by 2 workers (2180 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 4.12k samplesPerSecond , throughputPerWorker = 2.06k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.06881380 * 489; EvalErrorPrediction = 0.56237219 * 489; time = 0.2499s; samplesPerSecond = 1956.8
|
||||
MPI Rank 0: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.06291823 * 494; EvalErrorPrediction = 0.55668016 * 494; time = 0.2164s; samplesPerSecond = 2282.9
|
||||
MPI Rank 0: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 2.10126825 * 499; EvalErrorPrediction = 0.57915832 * 499; time = 0.2242s; samplesPerSecond = 2225.3
|
||||
MPI Rank 0: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.04906102 * 490; EvalErrorPrediction = 0.56938776 * 490; time = 0.2237s; samplesPerSecond = 2190.6
|
||||
MPI Rank 0: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.06881380 * 489; EvalClassificationError = 0.56237219 * 489; time = 0.2499s; samplesPerSecond = 1956.8
|
||||
MPI Rank 0: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.06291823 * 494; EvalClassificationError = 0.55668016 * 494; time = 0.2164s; samplesPerSecond = 2282.9
|
||||
MPI Rank 0: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 2.10126825 * 499; EvalClassificationError = 0.57915832 * 499; time = 0.2242s; samplesPerSecond = 2225.3
|
||||
MPI Rank 0: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.04906102 * 490; EvalClassificationError = 0.56938776 * 490; time = 0.2237s; samplesPerSecond = 2190.6
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.04-seconds latency this time; accumulated time on sync point = 0.09 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 1.03 seconds since last report (0.00 seconds on comm.); 4246 samples processed by 2 workers (2144 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 4.12k samplesPerSecond , throughputPerWorker = 2.06k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 1.92110713 * 497; EvalErrorPrediction = 0.50503018 * 497; time = 0.2847s; samplesPerSecond = 1745.5
|
||||
MPI Rank 0: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.11036477 * 492; EvalErrorPrediction = 0.57520325 * 492; time = 0.2308s; samplesPerSecond = 2132.1
|
||||
MPI Rank 0: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 1.90008474 * 508; EvalErrorPrediction = 0.53543307 * 508; time = 0.2381s; samplesPerSecond = 2134.0
|
||||
MPI Rank 0: 08/16/2016 10:08:30: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.96077526 * 503; EvalErrorPrediction = 0.52485089 * 503; time = 0.2373s; samplesPerSecond = 2119.7
|
||||
MPI Rank 0: 08/16/2016 10:08:30: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.06409953 * 470; EvalErrorPrediction = 0.57659574 * 470; time = 0.2194s; samplesPerSecond = 2142.0
|
||||
MPI Rank 0: 08/16/2016 10:08:30: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.89840653 * 494; EvalErrorPrediction = 0.53441296 * 494; time = 0.2267s; samplesPerSecond = 2179.1
|
||||
MPI Rank 0: 08/16/2016 10:08:30: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.97895657 * 503; EvalErrorPrediction = 0.52683897 * 503; time = 0.2352s; samplesPerSecond = 2138.8
|
||||
MPI Rank 0: 08/16/2016 10:08:30: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.94431225 * 487; EvalErrorPrediction = 0.53388090 * 487; time = 0.2178s; samplesPerSecond = 2236.5
|
||||
MPI Rank 0: 08/16/2016 10:08:31: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.94703738 * 516; EvalErrorPrediction = 0.50968992 * 516; time = 0.2328s; samplesPerSecond = 2216.6
|
||||
MPI Rank 0: 08/16/2016 10:08:31: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 1.95143083 * 494; EvalErrorPrediction = 0.55263158 * 494; time = 0.2281s; samplesPerSecond = 2165.4
|
||||
MPI Rank 0: 08/16/2016 10:08:31: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.99989131 * 510; EvalErrorPrediction = 0.54117647 * 510; time = 0.2424s; samplesPerSecond = 2104.0
|
||||
MPI Rank 0: 08/16/2016 10:08:31: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 2.00345753 * 497; EvalErrorPrediction = 0.54325956 * 497; time = 0.2268s; samplesPerSecond = 2191.0
|
||||
MPI Rank 0: 08/16/2016 10:08:32: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.89137230 * 490; EvalErrorPrediction = 0.54489796 * 490; time = 0.2227s; samplesPerSecond = 2200.2
|
||||
MPI Rank 0: 08/16/2016 10:08:32: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.94443698 * 482; EvalErrorPrediction = 0.52282158 * 482; time = 0.2314s; samplesPerSecond = 2082.8
|
||||
MPI Rank 0: 08/16/2016 10:08:32: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.97053909 * 342; EvalErrorPrediction = 0.51169591 * 342; time = 0.1615s; samplesPerSecond = 2117.3
|
||||
MPI Rank 0: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 1.92110713 * 497; EvalClassificationError = 0.50503018 * 497; time = 0.2847s; samplesPerSecond = 1745.5
|
||||
MPI Rank 0: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.11036477 * 492; EvalClassificationError = 0.57520325 * 492; time = 0.2308s; samplesPerSecond = 2132.1
|
||||
MPI Rank 0: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 1.90008474 * 508; EvalClassificationError = 0.53543307 * 508; time = 0.2381s; samplesPerSecond = 2134.0
|
||||
MPI Rank 0: 08/16/2016 10:08:30: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.96077526 * 503; EvalClassificationError = 0.52485089 * 503; time = 0.2373s; samplesPerSecond = 2119.7
|
||||
MPI Rank 0: 08/16/2016 10:08:30: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.06409953 * 470; EvalClassificationError = 0.57659574 * 470; time = 0.2194s; samplesPerSecond = 2142.0
|
||||
MPI Rank 0: 08/16/2016 10:08:30: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.89840653 * 494; EvalClassificationError = 0.53441296 * 494; time = 0.2267s; samplesPerSecond = 2179.1
|
||||
MPI Rank 0: 08/16/2016 10:08:30: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.97895657 * 503; EvalClassificationError = 0.52683897 * 503; time = 0.2352s; samplesPerSecond = 2138.8
|
||||
MPI Rank 0: 08/16/2016 10:08:30: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.94431225 * 487; EvalClassificationError = 0.53388090 * 487; time = 0.2178s; samplesPerSecond = 2236.5
|
||||
MPI Rank 0: 08/16/2016 10:08:31: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.94703738 * 516; EvalClassificationError = 0.50968992 * 516; time = 0.2328s; samplesPerSecond = 2216.6
|
||||
MPI Rank 0: 08/16/2016 10:08:31: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 1.95143083 * 494; EvalClassificationError = 0.55263158 * 494; time = 0.2281s; samplesPerSecond = 2165.4
|
||||
MPI Rank 0: 08/16/2016 10:08:31: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.99989131 * 510; EvalClassificationError = 0.54117647 * 510; time = 0.2424s; samplesPerSecond = 2104.0
|
||||
MPI Rank 0: 08/16/2016 10:08:31: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 2.00345753 * 497; EvalClassificationError = 0.54325956 * 497; time = 0.2268s; samplesPerSecond = 2191.0
|
||||
MPI Rank 0: 08/16/2016 10:08:32: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.89137230 * 490; EvalClassificationError = 0.54489796 * 490; time = 0.2227s; samplesPerSecond = 2200.2
|
||||
MPI Rank 0: 08/16/2016 10:08:32: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.94443698 * 482; EvalClassificationError = 0.52282158 * 482; time = 0.2314s; samplesPerSecond = 2082.8
|
||||
MPI Rank 0: 08/16/2016 10:08:32: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.97053909 * 342; EvalClassificationError = 0.51169591 * 342; time = 0.1615s; samplesPerSecond = 2117.3
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.09 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 3.17 seconds since last report (0.00 seconds on comm.); 7692 samples processed by 2 workers (6788 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 2.43k samplesPerSecond , throughputPerWorker = 1.21k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:32: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.03633283 * 20480; EvalErrorPrediction = 0.55610352 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=6.31975s
|
||||
MPI Rank 0: 08/16/2016 10:08:32: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.03633283 * 20480; EvalClassificationError = 0.55610352 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=6.31975s
|
||||
MPI Rank 0: 08/16/2016 10:08:32: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBM@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:32: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -718,26 +718,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:32: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:08:33: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.95205727 * 1942; EvalErrorPrediction = 0.53347065 * 1942; time = 0.8982s; samplesPerSecond = 2162.1
|
||||
MPI Rank 0: 08/16/2016 10:08:33: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.95205727 * 1942; EvalClassificationError = 0.53347065 * 1942; time = 0.8982s; samplesPerSecond = 2162.1
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 1.22 seconds since last report (0.01 seconds on comm.); 4885 samples processed by 2 workers (2592 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 4.01k samplesPerSecond , throughputPerWorker = 2.01k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:34: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.92969491 * 1909; EvalErrorPrediction = 0.54635935 * 1909; time = 0.8692s; samplesPerSecond = 2196.3
|
||||
MPI Rank 0: 08/16/2016 10:08:34: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.92969491 * 1909; EvalClassificationError = 0.54635935 * 1909; time = 0.8692s; samplesPerSecond = 2196.3
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 1.18 seconds since last report (0.01 seconds on comm.); 4826 samples processed by 2 workers (2577 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 4.08k samplesPerSecond , throughputPerWorker = 2.04k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:35: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.97762356 * 1987; EvalErrorPrediction = 0.55208858 * 1987; time = 0.9330s; samplesPerSecond = 2129.7
|
||||
MPI Rank 0: 08/16/2016 10:08:35: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.97762356 * 1987; EvalClassificationError = 0.55208858 * 1987; time = 0.9330s; samplesPerSecond = 2129.7
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 1.17 seconds since last report (0.01 seconds on comm.); 4903 samples processed by 2 workers (2577 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 4.21k samplesPerSecond , throughputPerWorker = 2.10k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:36: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.92634914 * 1908; EvalErrorPrediction = 0.54664570 * 1908; time = 0.8616s; samplesPerSecond = 2214.4
|
||||
MPI Rank 0: 08/16/2016 10:08:37: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.96187742 * 1905; EvalErrorPrediction = 0.55380577 * 1905; time = 0.8160s; samplesPerSecond = 2334.4
|
||||
MPI Rank 0: 08/16/2016 10:08:37: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.95001780 * 1913; EvalErrorPrediction = 0.54312598 * 1913; time = 0.8211s; samplesPerSecond = 2329.9
|
||||
MPI Rank 0: 08/16/2016 10:08:38: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.95834287 * 1225; EvalErrorPrediction = 0.55428571 * 1225; time = 0.5324s; samplesPerSecond = 2301.1
|
||||
MPI Rank 0: 08/16/2016 10:08:36: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.92634914 * 1908; EvalClassificationError = 0.54664570 * 1908; time = 0.8616s; samplesPerSecond = 2214.4
|
||||
MPI Rank 0: 08/16/2016 10:08:37: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.96187742 * 1905; EvalClassificationError = 0.55380577 * 1905; time = 0.8160s; samplesPerSecond = 2334.4
|
||||
MPI Rank 0: 08/16/2016 10:08:37: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.95001780 * 1913; EvalClassificationError = 0.54312598 * 1913; time = 0.8211s; samplesPerSecond = 2329.9
|
||||
MPI Rank 0: 08/16/2016 10:08:38: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.95834287 * 1225; EvalClassificationError = 0.55428571 * 1225; time = 0.5324s; samplesPerSecond = 2301.1
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 2.18 seconds since last report (0.00 seconds on comm.); 5866 samples processed by 2 workers (5043 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 2.69k samplesPerSecond , throughputPerWorker = 1.34k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:38: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.94581281 * 20480; EvalErrorPrediction = 0.54428711 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=5.75126s
|
||||
MPI Rank 0: 08/16/2016 10:08:38: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.94581281 * 20480; EvalClassificationError = 0.54428711 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=5.75126s
|
||||
MPI Rank 0: 08/16/2016 10:08:38: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBM@release_cpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:38: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -745,26 +745,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:38: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:08:39: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.88148355 * 1923; EvalErrorPrediction = 0.52002080 * 1923; time = 0.8512s; samplesPerSecond = 2259.3
|
||||
MPI Rank 0: 08/16/2016 10:08:39: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.88148355 * 1923; EvalClassificationError = 0.52002080 * 1923; time = 0.8512s; samplesPerSecond = 2259.3
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 1.15 seconds since last report (0.00 seconds on comm.); 4901 samples processed by 2 workers (2550 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 4.27k samplesPerSecond , throughputPerWorker = 2.14k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:40: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.88525866 * 1870; EvalErrorPrediction = 0.51764706 * 1870; time = 0.8251s; samplesPerSecond = 2266.3
|
||||
MPI Rank 0: 08/16/2016 10:08:40: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.88525866 * 1870; EvalClassificationError = 0.51764706 * 1870; time = 0.8251s; samplesPerSecond = 2266.3
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 1.11 seconds since last report (0.00 seconds on comm.); 4836 samples processed by 2 workers (2519 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 4.36k samplesPerSecond , throughputPerWorker = 2.18k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:41: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.89851024 * 1942; EvalErrorPrediction = 0.54376931 * 1942; time = 0.8657s; samplesPerSecond = 2243.1
|
||||
MPI Rank 0: 08/16/2016 10:08:41: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.89851024 * 1942; EvalClassificationError = 0.54376931 * 1942; time = 0.8657s; samplesPerSecond = 2243.1
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 1.11 seconds since last report (0.00 seconds on comm.); 4952 samples processed by 2 workers (2551 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 4.46k samplesPerSecond , throughputPerWorker = 2.23k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:41: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.86740185 * 1885; EvalErrorPrediction = 0.51405836 * 1885; time = 0.8208s; samplesPerSecond = 2296.6
|
||||
MPI Rank 0: 08/16/2016 10:08:42: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.89232216 * 1870; EvalErrorPrediction = 0.51229947 * 1870; time = 0.7965s; samplesPerSecond = 2347.8
|
||||
MPI Rank 0: 08/16/2016 10:08:43: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.87889109 * 1873; EvalErrorPrediction = 0.52162306 * 1873; time = 0.8236s; samplesPerSecond = 2274.2
|
||||
MPI Rank 0: 08/16/2016 10:08:43: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.91343765 * 1231; EvalErrorPrediction = 0.51827782 * 1231; time = 0.5226s; samplesPerSecond = 2355.5
|
||||
MPI Rank 0: 08/16/2016 10:08:41: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.86740185 * 1885; EvalClassificationError = 0.51405836 * 1885; time = 0.8208s; samplesPerSecond = 2296.6
|
||||
MPI Rank 0: 08/16/2016 10:08:42: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.89232216 * 1870; EvalClassificationError = 0.51229947 * 1870; time = 0.7965s; samplesPerSecond = 2347.8
|
||||
MPI Rank 0: 08/16/2016 10:08:43: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.87889109 * 1873; EvalClassificationError = 0.52162306 * 1873; time = 0.8236s; samplesPerSecond = 2274.2
|
||||
MPI Rank 0: 08/16/2016 10:08:43: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.91343765 * 1231; EvalClassificationError = 0.51827782 * 1231; time = 0.5226s; samplesPerSecond = 2355.5
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 2.16 seconds since last report (0.00 seconds on comm.); 5791 samples processed by 2 workers (4974 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 2.68k samplesPerSecond , throughputPerWorker = 1.34k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:44: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.89065307 * 20480; EvalErrorPrediction = 0.52431641 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=5.52517s
|
||||
MPI Rank 0: 08/16/2016 10:08:44: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.89065307 * 20480; EvalClassificationError = 0.52431641 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=5.52517s
|
||||
MPI Rank 0: 08/16/2016 10:08:44: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBM@release_cpu/models/cntkSpeech.dnn.4'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:44: Starting Epoch 5: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -772,26 +772,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 4: frames [81920..102400] (first utterance at frame 81920), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:44: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:08:44: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.92086178 * 1863; EvalErrorPrediction = 0.53140097 * 1863; time = 0.8247s; samplesPerSecond = 2258.9
|
||||
MPI Rank 0: 08/16/2016 10:08:44: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.92086178 * 1863; EvalClassificationError = 0.53140097 * 1863; time = 0.8247s; samplesPerSecond = 2258.9
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 1.12 seconds since last report (0.00 seconds on comm.); 4919 samples processed by 2 workers (2493 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 4.39k samplesPerSecond , throughputPerWorker = 2.19k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:45: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.85158217 * 1855; EvalErrorPrediction = 0.51967655 * 1855; time = 0.8172s; samplesPerSecond = 2269.8
|
||||
MPI Rank 0: 08/16/2016 10:08:45: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.85158217 * 1855; EvalClassificationError = 0.51967655 * 1855; time = 0.8172s; samplesPerSecond = 2269.8
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 1.10 seconds since last report (0.00 seconds on comm.); 4899 samples processed by 2 workers (2480 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 4.44k samplesPerSecond , throughputPerWorker = 2.22k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:46: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.84823089 * 1866; EvalErrorPrediction = 0.51768489 * 1866; time = 0.8309s; samplesPerSecond = 2245.7
|
||||
MPI Rank 0: 08/16/2016 10:08:46: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.84823089 * 1866; EvalClassificationError = 0.51768489 * 1866; time = 0.8309s; samplesPerSecond = 2245.7
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 1.08 seconds since last report (0.00 seconds on comm.); 4829 samples processed by 2 workers (2470 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 4.48k samplesPerSecond , throughputPerWorker = 2.24k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:47: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.86897551 * 1859; EvalErrorPrediction = 0.52124798 * 1859; time = 0.8262s; samplesPerSecond = 2250.1
|
||||
MPI Rank 0: 08/16/2016 10:08:48: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.80580101 * 1925; EvalErrorPrediction = 0.49506494 * 1925; time = 0.8559s; samplesPerSecond = 2249.0
|
||||
MPI Rank 0: 08/16/2016 10:08:49: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.83036468 * 1860; EvalErrorPrediction = 0.50322581 * 1860; time = 0.8128s; samplesPerSecond = 2288.5
|
||||
MPI Rank 0: 08/16/2016 10:08:49: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.84485058 * 1239; EvalErrorPrediction = 0.50524617 * 1239; time = 0.5158s; samplesPerSecond = 2402.2
|
||||
MPI Rank 0: 08/16/2016 10:08:47: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.86897551 * 1859; EvalClassificationError = 0.52124798 * 1859; time = 0.8262s; samplesPerSecond = 2250.1
|
||||
MPI Rank 0: 08/16/2016 10:08:48: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.80580101 * 1925; EvalClassificationError = 0.49506494 * 1925; time = 0.8559s; samplesPerSecond = 2249.0
|
||||
MPI Rank 0: 08/16/2016 10:08:49: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.83036468 * 1860; EvalClassificationError = 0.50322581 * 1860; time = 0.8128s; samplesPerSecond = 2288.5
|
||||
MPI Rank 0: 08/16/2016 10:08:49: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.84485058 * 1239; EvalClassificationError = 0.50524617 * 1239; time = 0.5158s; samplesPerSecond = 2402.2
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 2.20 seconds since last report (0.00 seconds on comm.); 5833 samples processed by 2 workers (5024 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 2.65k samplesPerSecond , throughputPerWorker = 1.33k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:49: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.87480299 * 20480; EvalErrorPrediction = 0.51845703 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-05; epochTime=5.50168s
|
||||
MPI Rank 0: 08/16/2016 10:08:49: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.87480299 * 20480; EvalClassificationError = 0.51845703 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-05; epochTime=5.50168s
|
||||
MPI Rank 0: 08/16/2016 10:08:49: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBM@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 10:08:49: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -844,7 +844,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -870,7 +870,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -952,7 +952,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -978,7 +978,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1066,7 +1066,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1092,7 +1092,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1191,7 +1191,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1220,7 +1220,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1243,7 +1243,7 @@ MPI Rank 1: 08/16/2016 10:08:13: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 10:08:13: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:13: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 10:08:13: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 10:08:13: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1296,237 +1296,237 @@ MPI Rank 1: 08/16/2016 10:08:15: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:15: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.62930672 * 192; EvalErrorPrediction = 0.90625000 * 192; time = 0.1036s; samplesPerSecond = 1853.9
|
||||
MPI Rank 1: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.39189508 * 192; EvalErrorPrediction = 0.88020833 * 192; time = 0.0964s; samplesPerSecond = 1991.0
|
||||
MPI Rank 1: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.73456191 * 192; EvalErrorPrediction = 0.92708333 * 192; time = 0.0941s; samplesPerSecond = 2039.3
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.22329144 * 192; EvalErrorPrediction = 0.91145833 * 192; time = 0.0982s; samplesPerSecond = 1956.2
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.44791490 * 192; EvalErrorPrediction = 0.90625000 * 192; time = 0.0957s; samplesPerSecond = 2005.9
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.62%]: CrossEntropyWithSoftmax = 4.37559721 * 192; EvalErrorPrediction = 0.93750000 * 192; time = 0.1013s; samplesPerSecond = 1895.8
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.24117585 * 192; EvalErrorPrediction = 0.94791667 * 192; time = 0.0952s; samplesPerSecond = 2017.7
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.05463916 * 192; EvalErrorPrediction = 0.89062500 * 192; time = 0.0945s; samplesPerSecond = 2031.3
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.98830106 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0963s; samplesPerSecond = 1994.7
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.84560829 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0961s; samplesPerSecond = 1998.1
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.81542483 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0988s; samplesPerSecond = 1943.6
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.75181899 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0964s; samplesPerSecond = 1991.1
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.74962983 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0974s; samplesPerSecond = 1970.7
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.12%]: CrossEntropyWithSoftmax = 3.70309729 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.1009s; samplesPerSecond = 1902.4
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89401044 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0961s; samplesPerSecond = 1998.0
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.78605845 * 192; EvalErrorPrediction = 0.83854167 * 192; time = 0.1010s; samplesPerSecond = 1901.6
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.82243497 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0983s; samplesPerSecond = 1954.1
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.91866810 * 192; EvalErrorPrediction = 0.85937500 * 192; time = 0.0998s; samplesPerSecond = 1923.3
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56637464 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0939s; samplesPerSecond = 2045.0
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.64635888 * 192; EvalErrorPrediction = 0.86458333 * 192; time = 0.0975s; samplesPerSecond = 1968.9
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.35602557 * 192; EvalErrorPrediction = 0.75520833 * 192; time = 0.0963s; samplesPerSecond = 1993.8
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.62%]: CrossEntropyWithSoftmax = 3.40627666 * 192; EvalErrorPrediction = 0.78645833 * 192; time = 0.0961s; samplesPerSecond = 1998.3
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.35477965 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0977s; samplesPerSecond = 1965.1
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.66184880 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0972s; samplesPerSecond = 1974.6
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.62630442 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0987s; samplesPerSecond = 1945.5
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.36376118 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0966s; samplesPerSecond = 1988.3
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51072509 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0986s; samplesPerSecond = 1946.3
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47385790 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0970s; samplesPerSecond = 1979.4
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.55227399 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0966s; samplesPerSecond = 1988.4
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.12%]: CrossEntropyWithSoftmax = 3.44007374 * 192; EvalErrorPrediction = 0.81250000 * 192; time = 0.0980s; samplesPerSecond = 1959.5
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.35761615 * 192; EvalErrorPrediction = 0.78645833 * 192; time = 0.0970s; samplesPerSecond = 1978.9
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55772259 * 192; EvalErrorPrediction = 0.81250000 * 192; time = 0.0950s; samplesPerSecond = 2020.2
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.33538937 * 192; EvalErrorPrediction = 0.81250000 * 192; time = 0.0989s; samplesPerSecond = 1940.5
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.51412750 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0955s; samplesPerSecond = 2010.3
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.56303150 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0985s; samplesPerSecond = 1949.1
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.48659009 * 192; EvalErrorPrediction = 0.83854167 * 192; time = 0.1005s; samplesPerSecond = 1910.8
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.33311671 * 192; EvalErrorPrediction = 0.80208333 * 192; time = 0.1032s; samplesPerSecond = 1861.2
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.62%]: CrossEntropyWithSoftmax = 3.28125589 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.1010s; samplesPerSecond = 1901.3
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.20377310 * 192; EvalErrorPrediction = 0.78645833 * 192; time = 0.0999s; samplesPerSecond = 1921.6
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27073629 * 192; EvalErrorPrediction = 0.78125000 * 192; time = 0.1001s; samplesPerSecond = 1917.8
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.55050536 * 192; EvalErrorPrediction = 0.83854167 * 192; time = 0.0965s; samplesPerSecond = 1988.9
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00419998 * 192; EvalErrorPrediction = 0.71354167 * 192; time = 0.0983s; samplesPerSecond = 1952.4
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.11647605 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0968s; samplesPerSecond = 1983.0
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.06195640 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.1003s; samplesPerSecond = 1913.8
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18915458 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0997s; samplesPerSecond = 1925.6
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.12%]: CrossEntropyWithSoftmax = 2.77654930 * 192; EvalErrorPrediction = 0.71354167 * 192; time = 0.1001s; samplesPerSecond = 1918.0
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17408570 * 192; EvalErrorPrediction = 0.83854167 * 192; time = 0.1029s; samplesPerSecond = 1865.8
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03022229 * 192; EvalErrorPrediction = 0.72916667 * 192; time = 0.0968s; samplesPerSecond = 1984.5
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.01692112 * 192; EvalErrorPrediction = 0.74479167 * 192; time = 0.1018s; samplesPerSecond = 1886.6
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79519565 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0973s; samplesPerSecond = 1973.2
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.03721970 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0993s; samplesPerSecond = 1933.8
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.16911303 * 192; EvalErrorPrediction = 0.78645833 * 192; time = 0.0959s; samplesPerSecond = 2001.5
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11747666 * 192; EvalErrorPrediction = 0.75000000 * 192; time = 0.0991s; samplesPerSecond = 1937.7
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.62%]: CrossEntropyWithSoftmax = 2.86999165 * 192; EvalErrorPrediction = 0.72395833 * 192; time = 0.0965s; samplesPerSecond = 1989.0
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.78866812 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.1008s; samplesPerSecond = 1905.0
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00684431 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.1012s; samplesPerSecond = 1897.6
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.72612865 * 192; EvalErrorPrediction = 0.68229167 * 192; time = 0.0984s; samplesPerSecond = 1951.5
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64981655 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0993s; samplesPerSecond = 1933.8
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71554609 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.1001s; samplesPerSecond = 1917.8
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.73031896 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0988s; samplesPerSecond = 1943.1
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.83317991 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.1010s; samplesPerSecond = 1901.2
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79636434 * 192; EvalErrorPrediction = 0.71354167 * 192; time = 0.0984s; samplesPerSecond = 1950.6
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.77040456 * 192; EvalErrorPrediction = 0.72395833 * 192; time = 0.0963s; samplesPerSecond = 1994.3
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.79003220 * 192; EvalErrorPrediction = 0.69791667 * 192; time = 0.0949s; samplesPerSecond = 2024.2
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.54457731 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0994s; samplesPerSecond = 1931.6
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97922473 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0961s; samplesPerSecond = 1998.9
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.62424232 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0975s; samplesPerSecond = 1969.5
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52182003 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.1005s; samplesPerSecond = 1911.0
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.60628500 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0982s; samplesPerSecond = 1954.9
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.62%]: CrossEntropyWithSoftmax = 2.57508979 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0977s; samplesPerSecond = 1965.6
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.57480702 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0959s; samplesPerSecond = 2002.5
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.47514723 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0999s; samplesPerSecond = 1922.5
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.73668923 * 192; EvalErrorPrediction = 0.67187500 * 192; time = 0.0988s; samplesPerSecond = 1943.1
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.47244978 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0970s; samplesPerSecond = 1979.9
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.43386779 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0963s; samplesPerSecond = 1994.6
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.71554733 * 192; EvalErrorPrediction = 0.68750000 * 192; time = 0.0991s; samplesPerSecond = 1937.8
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.54334345 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0987s; samplesPerSecond = 1946.0
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.12%]: CrossEntropyWithSoftmax = 2.97324196 * 192; EvalErrorPrediction = 0.76041667 * 192; time = 0.0972s; samplesPerSecond = 1974.3
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.33516568 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0975s; samplesPerSecond = 1968.9
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.23213800 * 192; EvalErrorPrediction = 0.60937500 * 192; time = 0.0983s; samplesPerSecond = 1953.9
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.29168780 * 192; EvalErrorPrediction = 0.57812500 * 192; time = 0.0970s; samplesPerSecond = 1979.5
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.68110845 * 192; EvalErrorPrediction = 0.70312500 * 192; time = 0.0987s; samplesPerSecond = 1945.1
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.43379863 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.1000s; samplesPerSecond = 1920.3
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.48936113 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0988s; samplesPerSecond = 1942.7
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.37855205 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0967s; samplesPerSecond = 1985.8
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.62%]: CrossEntropyWithSoftmax = 2.35951804 * 192; EvalErrorPrediction = 0.57812500 * 192; time = 0.1001s; samplesPerSecond = 1917.8
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.32828318 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0946s; samplesPerSecond = 2029.1
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.27181872 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0976s; samplesPerSecond = 1968.0
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.07018832 * 192; EvalErrorPrediction = 0.50520833 * 192; time = 0.0965s; samplesPerSecond = 1989.1
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.40647801 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0989s; samplesPerSecond = 1940.4
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.51858879 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0971s; samplesPerSecond = 1977.2
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33656120 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0978s; samplesPerSecond = 1964.0
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.60987825 * 192; EvalErrorPrediction = 0.69791667 * 192; time = 0.0927s; samplesPerSecond = 2070.7
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.12%]: CrossEntropyWithSoftmax = 2.46713772 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.1014s; samplesPerSecond = 1893.5
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30355405 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0997s; samplesPerSecond = 1925.0
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.19148527 * 192; EvalErrorPrediction = 0.59895833 * 192; time = 0.0978s; samplesPerSecond = 1962.5
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.25368531 * 192; EvalErrorPrediction = 0.56250000 * 192; time = 0.0981s; samplesPerSecond = 1958.0
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.10763074 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0981s; samplesPerSecond = 1958.1
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.30064187 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0966s; samplesPerSecond = 1986.7
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.27695669 * 192; EvalErrorPrediction = 0.59375000 * 192; time = 0.0991s; samplesPerSecond = 1938.2
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.13317518 * 192; EvalErrorPrediction = 0.56250000 * 192; time = 0.0993s; samplesPerSecond = 1933.9
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.62%]: CrossEntropyWithSoftmax = 2.23952900 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0967s; samplesPerSecond = 1984.8
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.13404008 * 192; EvalErrorPrediction = 0.54687500 * 192; time = 0.0990s; samplesPerSecond = 1939.1
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.42845170 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0999s; samplesPerSecond = 1922.8
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.15843921 * 192; EvalErrorPrediction = 0.60937500 * 192; time = 0.1015s; samplesPerSecond = 1891.9
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.20218732 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0998s; samplesPerSecond = 1924.1
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=10.4855s
|
||||
MPI Rank 1: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.62930672 * 192; EvalClassificationError = 0.90625000 * 192; time = 0.1036s; samplesPerSecond = 1853.9
|
||||
MPI Rank 1: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.39189508 * 192; EvalClassificationError = 0.88020833 * 192; time = 0.0964s; samplesPerSecond = 1991.0
|
||||
MPI Rank 1: 08/16/2016 10:08:15: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.73456191 * 192; EvalClassificationError = 0.92708333 * 192; time = 0.0941s; samplesPerSecond = 2039.3
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.22329144 * 192; EvalClassificationError = 0.91145833 * 192; time = 0.0982s; samplesPerSecond = 1956.2
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.44791490 * 192; EvalClassificationError = 0.90625000 * 192; time = 0.0957s; samplesPerSecond = 2005.9
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.62%]: CrossEntropyWithSoftmax = 4.37559721 * 192; EvalClassificationError = 0.93750000 * 192; time = 0.1013s; samplesPerSecond = 1895.8
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.24117585 * 192; EvalClassificationError = 0.94791667 * 192; time = 0.0952s; samplesPerSecond = 2017.7
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.05463916 * 192; EvalClassificationError = 0.89062500 * 192; time = 0.0945s; samplesPerSecond = 2031.3
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.98830106 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0963s; samplesPerSecond = 1994.7
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.84560829 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0961s; samplesPerSecond = 1998.1
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.81542483 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0988s; samplesPerSecond = 1943.6
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.75181899 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0964s; samplesPerSecond = 1991.1
|
||||
MPI Rank 1: 08/16/2016 10:08:16: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.74962983 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0974s; samplesPerSecond = 1970.7
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.12%]: CrossEntropyWithSoftmax = 3.70309729 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.1009s; samplesPerSecond = 1902.4
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89401044 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0961s; samplesPerSecond = 1998.0
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.78605845 * 192; EvalClassificationError = 0.83854167 * 192; time = 0.1010s; samplesPerSecond = 1901.6
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.82243497 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0983s; samplesPerSecond = 1954.1
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.91866810 * 192; EvalClassificationError = 0.85937500 * 192; time = 0.0998s; samplesPerSecond = 1923.3
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56637464 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0939s; samplesPerSecond = 2045.0
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.64635888 * 192; EvalClassificationError = 0.86458333 * 192; time = 0.0975s; samplesPerSecond = 1968.9
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.35602557 * 192; EvalClassificationError = 0.75520833 * 192; time = 0.0963s; samplesPerSecond = 1993.8
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.62%]: CrossEntropyWithSoftmax = 3.40627666 * 192; EvalClassificationError = 0.78645833 * 192; time = 0.0961s; samplesPerSecond = 1998.3
|
||||
MPI Rank 1: 08/16/2016 10:08:17: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.35477965 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0977s; samplesPerSecond = 1965.1
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.66184880 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0972s; samplesPerSecond = 1974.6
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.62630442 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0987s; samplesPerSecond = 1945.5
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.36376118 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0966s; samplesPerSecond = 1988.3
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51072509 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0986s; samplesPerSecond = 1946.3
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47385790 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0970s; samplesPerSecond = 1979.4
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.55227399 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0966s; samplesPerSecond = 1988.4
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.12%]: CrossEntropyWithSoftmax = 3.44007374 * 192; EvalClassificationError = 0.81250000 * 192; time = 0.0980s; samplesPerSecond = 1959.5
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.35761615 * 192; EvalClassificationError = 0.78645833 * 192; time = 0.0970s; samplesPerSecond = 1978.9
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55772259 * 192; EvalClassificationError = 0.81250000 * 192; time = 0.0950s; samplesPerSecond = 2020.2
|
||||
MPI Rank 1: 08/16/2016 10:08:18: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.33538937 * 192; EvalClassificationError = 0.81250000 * 192; time = 0.0989s; samplesPerSecond = 1940.5
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.51412750 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0955s; samplesPerSecond = 2010.3
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.56303150 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0985s; samplesPerSecond = 1949.1
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.48659009 * 192; EvalClassificationError = 0.83854167 * 192; time = 0.1005s; samplesPerSecond = 1910.8
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.33311671 * 192; EvalClassificationError = 0.80208333 * 192; time = 0.1032s; samplesPerSecond = 1861.2
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.62%]: CrossEntropyWithSoftmax = 3.28125589 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.1010s; samplesPerSecond = 1901.3
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.20377310 * 192; EvalClassificationError = 0.78645833 * 192; time = 0.0999s; samplesPerSecond = 1921.6
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27073629 * 192; EvalClassificationError = 0.78125000 * 192; time = 0.1001s; samplesPerSecond = 1917.8
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.55050536 * 192; EvalClassificationError = 0.83854167 * 192; time = 0.0965s; samplesPerSecond = 1988.9
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00419998 * 192; EvalClassificationError = 0.71354167 * 192; time = 0.0983s; samplesPerSecond = 1952.4
|
||||
MPI Rank 1: 08/16/2016 10:08:19: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.11647605 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0968s; samplesPerSecond = 1983.0
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.06195640 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.1003s; samplesPerSecond = 1913.8
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18915458 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0997s; samplesPerSecond = 1925.6
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.12%]: CrossEntropyWithSoftmax = 2.77654930 * 192; EvalClassificationError = 0.71354167 * 192; time = 0.1001s; samplesPerSecond = 1918.0
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17408570 * 192; EvalClassificationError = 0.83854167 * 192; time = 0.1029s; samplesPerSecond = 1865.8
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03022229 * 192; EvalClassificationError = 0.72916667 * 192; time = 0.0968s; samplesPerSecond = 1984.5
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.01692112 * 192; EvalClassificationError = 0.74479167 * 192; time = 0.1018s; samplesPerSecond = 1886.6
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79519565 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0973s; samplesPerSecond = 1973.2
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.03721970 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0993s; samplesPerSecond = 1933.8
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.16911303 * 192; EvalClassificationError = 0.78645833 * 192; time = 0.0959s; samplesPerSecond = 2001.5
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11747666 * 192; EvalClassificationError = 0.75000000 * 192; time = 0.0991s; samplesPerSecond = 1937.7
|
||||
MPI Rank 1: 08/16/2016 10:08:20: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.62%]: CrossEntropyWithSoftmax = 2.86999165 * 192; EvalClassificationError = 0.72395833 * 192; time = 0.0965s; samplesPerSecond = 1989.0
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.78866812 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.1008s; samplesPerSecond = 1905.0
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00684431 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.1012s; samplesPerSecond = 1897.6
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.72612865 * 192; EvalClassificationError = 0.68229167 * 192; time = 0.0984s; samplesPerSecond = 1951.5
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64981655 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0993s; samplesPerSecond = 1933.8
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71554609 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.1001s; samplesPerSecond = 1917.8
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.73031896 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0988s; samplesPerSecond = 1943.1
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.83317991 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.1010s; samplesPerSecond = 1901.2
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79636434 * 192; EvalClassificationError = 0.71354167 * 192; time = 0.0984s; samplesPerSecond = 1950.6
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.77040456 * 192; EvalClassificationError = 0.72395833 * 192; time = 0.0963s; samplesPerSecond = 1994.3
|
||||
MPI Rank 1: 08/16/2016 10:08:21: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.79003220 * 192; EvalClassificationError = 0.69791667 * 192; time = 0.0949s; samplesPerSecond = 2024.2
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.54457731 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0994s; samplesPerSecond = 1931.6
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97922473 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0961s; samplesPerSecond = 1998.9
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.62424232 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0975s; samplesPerSecond = 1969.5
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52182003 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.1005s; samplesPerSecond = 1911.0
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.60628500 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0982s; samplesPerSecond = 1954.9
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.62%]: CrossEntropyWithSoftmax = 2.57508979 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0977s; samplesPerSecond = 1965.6
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.57480702 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0959s; samplesPerSecond = 2002.5
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.47514723 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0999s; samplesPerSecond = 1922.5
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.73668923 * 192; EvalClassificationError = 0.67187500 * 192; time = 0.0988s; samplesPerSecond = 1943.1
|
||||
MPI Rank 1: 08/16/2016 10:08:22: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.47244978 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0970s; samplesPerSecond = 1979.9
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.43386779 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0963s; samplesPerSecond = 1994.6
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.71554733 * 192; EvalClassificationError = 0.68750000 * 192; time = 0.0991s; samplesPerSecond = 1937.8
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.54334345 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0987s; samplesPerSecond = 1946.0
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.12%]: CrossEntropyWithSoftmax = 2.97324196 * 192; EvalClassificationError = 0.76041667 * 192; time = 0.0972s; samplesPerSecond = 1974.3
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.33516568 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0975s; samplesPerSecond = 1968.9
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.23213800 * 192; EvalClassificationError = 0.60937500 * 192; time = 0.0983s; samplesPerSecond = 1953.9
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.29168780 * 192; EvalClassificationError = 0.57812500 * 192; time = 0.0970s; samplesPerSecond = 1979.5
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.68110845 * 192; EvalClassificationError = 0.70312500 * 192; time = 0.0987s; samplesPerSecond = 1945.1
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.43379863 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.1000s; samplesPerSecond = 1920.3
|
||||
MPI Rank 1: 08/16/2016 10:08:23: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.48936113 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0988s; samplesPerSecond = 1942.7
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.37855205 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0967s; samplesPerSecond = 1985.8
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.62%]: CrossEntropyWithSoftmax = 2.35951804 * 192; EvalClassificationError = 0.57812500 * 192; time = 0.1001s; samplesPerSecond = 1917.8
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.32828318 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0946s; samplesPerSecond = 2029.1
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.27181872 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0976s; samplesPerSecond = 1968.0
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.07018832 * 192; EvalClassificationError = 0.50520833 * 192; time = 0.0965s; samplesPerSecond = 1989.1
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.40647801 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0989s; samplesPerSecond = 1940.4
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.51858879 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0971s; samplesPerSecond = 1977.2
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33656120 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0978s; samplesPerSecond = 1964.0
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.60987825 * 192; EvalClassificationError = 0.69791667 * 192; time = 0.0927s; samplesPerSecond = 2070.7
|
||||
MPI Rank 1: 08/16/2016 10:08:24: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.12%]: CrossEntropyWithSoftmax = 2.46713772 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.1014s; samplesPerSecond = 1893.5
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30355405 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0997s; samplesPerSecond = 1925.0
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.19148527 * 192; EvalClassificationError = 0.59895833 * 192; time = 0.0978s; samplesPerSecond = 1962.5
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.25368531 * 192; EvalClassificationError = 0.56250000 * 192; time = 0.0981s; samplesPerSecond = 1958.0
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.10763074 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0981s; samplesPerSecond = 1958.1
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.30064187 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0966s; samplesPerSecond = 1986.7
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.27695669 * 192; EvalClassificationError = 0.59375000 * 192; time = 0.0991s; samplesPerSecond = 1938.2
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.13317518 * 192; EvalClassificationError = 0.56250000 * 192; time = 0.0993s; samplesPerSecond = 1933.9
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.62%]: CrossEntropyWithSoftmax = 2.23952900 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0967s; samplesPerSecond = 1984.8
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.13404008 * 192; EvalClassificationError = 0.54687500 * 192; time = 0.0990s; samplesPerSecond = 1939.1
|
||||
MPI Rank 1: 08/16/2016 10:08:25: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.42845170 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0999s; samplesPerSecond = 1922.8
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.15843921 * 192; EvalClassificationError = 0.60937500 * 192; time = 0.1015s; samplesPerSecond = 1891.9
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.20218732 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0998s; samplesPerSecond = 1924.1
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=10.4855s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.17115762 * 260; EvalErrorPrediction = 0.59230769 * 260; time = 0.1338s; samplesPerSecond = 1943.1
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.32119931 * 276; EvalErrorPrediction = 0.68840580 * 276; time = 0.1411s; samplesPerSecond = 1956.1
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.22885229 * 280; EvalErrorPrediction = 0.65714286 * 280; time = 0.1360s; samplesPerSecond = 2058.9
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.18860582 * 241; EvalErrorPrediction = 0.64730290 * 241; time = 0.1221s; samplesPerSecond = 1973.9
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.04470099 * 295; EvalErrorPrediction = 0.54237288 * 295; time = 0.1513s; samplesPerSecond = 1949.6
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.36015794 * 257; EvalErrorPrediction = 0.62256809 * 257; time = 0.1256s; samplesPerSecond = 2046.4
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.01859796 * 262; EvalErrorPrediction = 0.58015267 * 262; time = 0.1299s; samplesPerSecond = 2017.7
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.17115762 * 260; EvalClassificationError = 0.59230769 * 260; time = 0.1338s; samplesPerSecond = 1943.1
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.32119931 * 276; EvalClassificationError = 0.68840580 * 276; time = 0.1411s; samplesPerSecond = 1956.1
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.22885229 * 280; EvalClassificationError = 0.65714286 * 280; time = 0.1360s; samplesPerSecond = 2058.9
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.18860582 * 241; EvalClassificationError = 0.64730290 * 241; time = 0.1221s; samplesPerSecond = 1973.9
|
||||
MPI Rank 1: 08/16/2016 10:08:26: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.04470099 * 295; EvalClassificationError = 0.54237288 * 295; time = 0.1513s; samplesPerSecond = 1949.6
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.36015794 * 257; EvalClassificationError = 0.62256809 * 257; time = 0.1256s; samplesPerSecond = 2046.4
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.01859796 * 262; EvalClassificationError = 0.58015267 * 262; time = 0.1299s; samplesPerSecond = 2017.7
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 1.09 seconds since last report (0.00 seconds on comm.); 4289 samples processed by 2 workers (2126 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 3.94k samplesPerSecond , throughputPerWorker = 1.97k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.01948678 * 255; EvalErrorPrediction = 0.52549020 * 255; time = 0.1428s; samplesPerSecond = 1786.1
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.06416722 * 279; EvalErrorPrediction = 0.59139785 * 279; time = 0.1376s; samplesPerSecond = 2028.1
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.03645193 * 274; EvalErrorPrediction = 0.53284672 * 274; time = 0.1331s; samplesPerSecond = 2059.2
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 1.94443177 * 269; EvalErrorPrediction = 0.55390335 * 269; time = 0.1324s; samplesPerSecond = 2031.3
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.02326632 * 278; EvalErrorPrediction = 0.61151079 * 278; time = 0.1344s; samplesPerSecond = 2068.3
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 2.10777520 * 271; EvalErrorPrediction = 0.53874539 * 271; time = 0.1299s; samplesPerSecond = 2085.9
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.19140851 * 276; EvalErrorPrediction = 0.58695652 * 276; time = 0.1344s; samplesPerSecond = 2053.2
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 2.14462135 * 260; EvalErrorPrediction = 0.58461538 * 260; time = 0.1293s; samplesPerSecond = 2010.5
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.01948678 * 255; EvalClassificationError = 0.52549020 * 255; time = 0.1428s; samplesPerSecond = 1786.1
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.06416722 * 279; EvalClassificationError = 0.59139785 * 279; time = 0.1376s; samplesPerSecond = 2028.1
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.03645193 * 274; EvalClassificationError = 0.53284672 * 274; time = 0.1331s; samplesPerSecond = 2059.2
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 1.94443177 * 269; EvalClassificationError = 0.55390335 * 269; time = 0.1324s; samplesPerSecond = 2031.3
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.02326632 * 278; EvalClassificationError = 0.61151079 * 278; time = 0.1344s; samplesPerSecond = 2068.3
|
||||
MPI Rank 1: 08/16/2016 10:08:27: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 2.10777520 * 271; EvalClassificationError = 0.53874539 * 271; time = 0.1299s; samplesPerSecond = 2085.9
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.19140851 * 276; EvalClassificationError = 0.58695652 * 276; time = 0.1344s; samplesPerSecond = 2053.2
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 2.14462135 * 260; EvalClassificationError = 0.58461538 * 260; time = 0.1293s; samplesPerSecond = 2010.5
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 1.03 seconds since last report (0.00 seconds on comm.); 4253 samples processed by 2 workers (2073 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 4.12k samplesPerSecond , throughputPerWorker = 2.06k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.96686419 * 265; EvalErrorPrediction = 0.55849057 * 265; time = 0.1499s; samplesPerSecond = 1768.1
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.12827703 * 298; EvalErrorPrediction = 0.57718121 * 298; time = 0.1415s; samplesPerSecond = 2105.4
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.95403115 * 274; EvalErrorPrediction = 0.51459854 * 274; time = 0.1325s; samplesPerSecond = 2068.5
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.95937773 * 265; EvalErrorPrediction = 0.53962264 * 265; time = 0.1311s; samplesPerSecond = 2022.1
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.84345811 * 281; EvalErrorPrediction = 0.54448399 * 281; time = 0.1372s; samplesPerSecond = 2047.9
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.93656418 * 252; EvalErrorPrediction = 0.51190476 * 252; time = 0.1215s; samplesPerSecond = 2073.8
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 2.02824763 * 274; EvalErrorPrediction = 0.51824818 * 274; time = 0.1318s; samplesPerSecond = 2079.3
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.83079859 * 258; EvalErrorPrediction = 0.49224806 * 258; time = 0.1253s; samplesPerSecond = 2059.4
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.96686419 * 265; EvalClassificationError = 0.55849057 * 265; time = 0.1499s; samplesPerSecond = 1768.1
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.12827703 * 298; EvalClassificationError = 0.57718121 * 298; time = 0.1415s; samplesPerSecond = 2105.4
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.95403115 * 274; EvalClassificationError = 0.51459854 * 274; time = 0.1325s; samplesPerSecond = 2068.5
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.95937773 * 265; EvalClassificationError = 0.53962264 * 265; time = 0.1311s; samplesPerSecond = 2022.1
|
||||
MPI Rank 1: 08/16/2016 10:08:28: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.84345811 * 281; EvalClassificationError = 0.54448399 * 281; time = 0.1372s; samplesPerSecond = 2047.9
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.93656418 * 252; EvalClassificationError = 0.51190476 * 252; time = 0.1215s; samplesPerSecond = 2073.8
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 2.02824763 * 274; EvalClassificationError = 0.51824818 * 274; time = 0.1318s; samplesPerSecond = 2079.3
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.83079859 * 258; EvalClassificationError = 0.49224806 * 258; time = 0.1253s; samplesPerSecond = 2059.4
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 1.03 seconds since last report (0.00 seconds on comm.); 4246 samples processed by 2 workers (2102 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 4.12k samplesPerSecond , throughputPerWorker = 2.06k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 1.95439604 * 271; EvalErrorPrediction = 0.54612546 * 271; time = 0.1415s; samplesPerSecond = 1915.5
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.97957556 * 278; EvalErrorPrediction = 0.52877698 * 278; time = 0.1328s; samplesPerSecond = 2094.0
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.92686640 * 286; EvalErrorPrediction = 0.54895105 * 286; time = 0.1394s; samplesPerSecond = 2051.2
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.94069763 * 170; EvalErrorPrediction = 0.51764706 * 170; time = 0.0831s; samplesPerSecond = 2044.5
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 1.95439604 * 271; EvalClassificationError = 0.54612546 * 271; time = 0.1415s; samplesPerSecond = 1915.5
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.97957556 * 278; EvalClassificationError = 0.52877698 * 278; time = 0.1328s; samplesPerSecond = 2094.0
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.92686640 * 286; EvalClassificationError = 0.54895105 * 286; time = 0.1394s; samplesPerSecond = 2051.2
|
||||
MPI Rank 1: 08/16/2016 10:08:29: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.94069763 * 170; EvalClassificationError = 0.51764706 * 170; time = 0.0831s; samplesPerSecond = 2044.5
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.56-seconds latency this time; accumulated time on sync point = 0.57 seconds , average latency = 0.14 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 3.17 seconds since last report (2.15 seconds on comm.); 7692 samples processed by 2 workers (904 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 2.43k samplesPerSecond , throughputPerWorker = 1.21k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:32: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.03633283 * 20480; EvalErrorPrediction = 0.55610352 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=6.31975s
|
||||
MPI Rank 1: 08/16/2016 10:08:32: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.03633283 * 20480; EvalClassificationError = 0.55610352 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=6.31975s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:32: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:32: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:08:33: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.93152418 * 1130; EvalErrorPrediction = 0.52654867 * 1130; time = 0.5272s; samplesPerSecond = 2143.4
|
||||
MPI Rank 1: 08/16/2016 10:08:33: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.93152418 * 1130; EvalClassificationError = 0.52654867 * 1130; time = 0.5272s; samplesPerSecond = 2143.4
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.15-seconds latency this time; accumulated time on sync point = 0.15 seconds , average latency = 0.15 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 1.22 seconds since last report (0.00 seconds on comm.); 4885 samples processed by 2 workers (2293 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 4.01k samplesPerSecond , throughputPerWorker = 2.01k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:33: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.86373194 * 1163; EvalErrorPrediction = 0.52278590 * 1163; time = 0.6866s; samplesPerSecond = 1693.8
|
||||
MPI Rank 1: 08/16/2016 10:08:34: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.97430683 * 1085; EvalErrorPrediction = 0.52718894 * 1085; time = 0.4813s; samplesPerSecond = 2254.3
|
||||
MPI Rank 1: 08/16/2016 10:08:33: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.86373194 * 1163; EvalClassificationError = 0.52278590 * 1163; time = 0.6866s; samplesPerSecond = 1693.8
|
||||
MPI Rank 1: 08/16/2016 10:08:34: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.97430683 * 1085; EvalClassificationError = 0.52718894 * 1085; time = 0.4813s; samplesPerSecond = 2254.3
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.20-seconds latency this time; accumulated time on sync point = 0.35 seconds , average latency = 0.17 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 1.18 seconds since last report (0.00 seconds on comm.); 4826 samples processed by 2 workers (2249 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 4.08k samplesPerSecond , throughputPerWorker = 2.04k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:35: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.92171339 * 1164; EvalErrorPrediction = 0.55670103 * 1164; time = 0.7021s; samplesPerSecond = 1657.9
|
||||
MPI Rank 1: 08/16/2016 10:08:35: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.99473898 * 1167; EvalErrorPrediction = 0.56726650 * 1167; time = 0.5007s; samplesPerSecond = 2330.8
|
||||
MPI Rank 1: 08/16/2016 10:08:35: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.92171339 * 1164; EvalClassificationError = 0.55670103 * 1164; time = 0.7021s; samplesPerSecond = 1657.9
|
||||
MPI Rank 1: 08/16/2016 10:08:35: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.99473898 * 1167; EvalClassificationError = 0.56726650 * 1167; time = 0.5007s; samplesPerSecond = 2330.8
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.14-seconds latency this time; accumulated time on sync point = 0.48 seconds , average latency = 0.16 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 1.17 seconds since last report (0.00 seconds on comm.); 4903 samples processed by 2 workers (2326 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 4.21k samplesPerSecond , throughputPerWorker = 2.10k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:36: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.91433800 * 1159; EvalErrorPrediction = 0.53580673 * 1159; time = 0.6646s; samplesPerSecond = 1743.9
|
||||
MPI Rank 1: 08/16/2016 10:08:36: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.97825377 * 823; EvalErrorPrediction = 0.54556501 * 823; time = 0.3631s; samplesPerSecond = 2266.5
|
||||
MPI Rank 1: 08/16/2016 10:08:36: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.91433800 * 1159; EvalClassificationError = 0.53580673 * 1159; time = 0.6646s; samplesPerSecond = 1743.9
|
||||
MPI Rank 1: 08/16/2016 10:08:36: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.97825377 * 823; EvalClassificationError = 0.54556501 * 823; time = 0.3631s; samplesPerSecond = 2266.5
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.72-seconds latency this time; accumulated time on sync point = 1.20 seconds , average latency = 0.30 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 2.18 seconds since last report (1.09 seconds on comm.); 5866 samples processed by 2 workers (823 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 2.69k samplesPerSecond , throughputPerWorker = 1.34k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:38: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.94581281 * 20480; EvalErrorPrediction = 0.54428711 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=5.75127s
|
||||
MPI Rank 1: 08/16/2016 10:08:38: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.94581281 * 20480; EvalClassificationError = 0.54428711 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=5.75127s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:38: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:38: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:08:38: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.88377244 * 1149; EvalErrorPrediction = 0.52306353 * 1149; time = 0.5089s; samplesPerSecond = 2257.6
|
||||
MPI Rank 1: 08/16/2016 10:08:38: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.88377244 * 1149; EvalClassificationError = 0.52306353 * 1149; time = 0.5089s; samplesPerSecond = 2257.6
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.08-seconds latency this time; accumulated time on sync point = 0.08 seconds , average latency = 0.08 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 1.15 seconds since last report (0.00 seconds on comm.); 4901 samples processed by 2 workers (2351 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 4.27k samplesPerSecond , throughputPerWorker = 2.14k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:39: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.93460753 * 1202; EvalErrorPrediction = 0.53577371 * 1202; time = 0.6336s; samplesPerSecond = 1897.1
|
||||
MPI Rank 1: 08/16/2016 10:08:40: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.88934249 * 1130; EvalErrorPrediction = 0.52212389 * 1130; time = 0.4886s; samplesPerSecond = 2312.9
|
||||
MPI Rank 1: 08/16/2016 10:08:39: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.93460753 * 1202; EvalClassificationError = 0.53577371 * 1202; time = 0.6336s; samplesPerSecond = 1897.1
|
||||
MPI Rank 1: 08/16/2016 10:08:40: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.88934249 * 1130; EvalClassificationError = 0.52212389 * 1130; time = 0.4886s; samplesPerSecond = 2312.9
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.09-seconds latency this time; accumulated time on sync point = 0.17 seconds , average latency = 0.08 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 1.11 seconds since last report (0.00 seconds on comm.); 4836 samples processed by 2 workers (2317 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 4.36k samplesPerSecond , throughputPerWorker = 2.18k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:40: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.84588720 * 1187; EvalErrorPrediction = 0.50716091 * 1187; time = 0.6214s; samplesPerSecond = 1910.3
|
||||
MPI Rank 1: 08/16/2016 10:08:41: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.92321892 * 1202; EvalErrorPrediction = 0.54409318 * 1202; time = 0.5341s; samplesPerSecond = 2250.7
|
||||
MPI Rank 1: 08/16/2016 10:08:40: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.84588720 * 1187; EvalClassificationError = 0.50716091 * 1187; time = 0.6214s; samplesPerSecond = 1910.3
|
||||
MPI Rank 1: 08/16/2016 10:08:41: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.92321892 * 1202; EvalClassificationError = 0.54409318 * 1202; time = 0.5341s; samplesPerSecond = 2250.7
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.06-seconds latency this time; accumulated time on sync point = 0.23 seconds , average latency = 0.08 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 1.11 seconds since last report (0.00 seconds on comm.); 4952 samples processed by 2 workers (2401 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 4.46k samplesPerSecond , throughputPerWorker = 2.23k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:41: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.89626910 * 1199; EvalErrorPrediction = 0.53878232 * 1199; time = 0.5769s; samplesPerSecond = 2078.5
|
||||
MPI Rank 1: 08/16/2016 10:08:42: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.90407355 * 817; EvalErrorPrediction = 0.53243574 * 817; time = 0.3636s; samplesPerSecond = 2246.8
|
||||
MPI Rank 1: 08/16/2016 10:08:41: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.89626910 * 1199; EvalClassificationError = 0.53878232 * 1199; time = 0.5769s; samplesPerSecond = 2078.5
|
||||
MPI Rank 1: 08/16/2016 10:08:42: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.90407355 * 817; EvalClassificationError = 0.53243574 * 817; time = 0.3636s; samplesPerSecond = 2246.8
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.71-seconds latency this time; accumulated time on sync point = 0.94 seconds , average latency = 0.23 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 2.16 seconds since last report (1.08 seconds on comm.); 5791 samples processed by 2 workers (817 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 2.68k samplesPerSecond , throughputPerWorker = 1.34k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:44: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.89065307 * 20480; EvalErrorPrediction = 0.52431641 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=5.52517s
|
||||
MPI Rank 1: 08/16/2016 10:08:44: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.89065307 * 20480; EvalClassificationError = 0.52431641 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=5.52517s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:44: Starting Epoch 5: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 4: frames [81920..102400] (first utterance at frame 81920), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:44: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:08:44: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.85747922 * 1209; EvalErrorPrediction = 0.50868486 * 1209; time = 0.5400s; samplesPerSecond = 2239.0
|
||||
MPI Rank 1: 08/16/2016 10:08:44: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.85747922 * 1209; EvalClassificationError = 0.50868486 * 1209; time = 0.5400s; samplesPerSecond = 2239.0
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.04-seconds latency this time; accumulated time on sync point = 0.04 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 1.12 seconds since last report (0.00 seconds on comm.); 4919 samples processed by 2 workers (2426 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 4.39k samplesPerSecond , throughputPerWorker = 2.19k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:45: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.92784049 * 1217; EvalErrorPrediction = 0.52670501 * 1217; time = 0.5780s; samplesPerSecond = 2105.4
|
||||
MPI Rank 1: 08/16/2016 10:08:45: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.88585235 * 1206; EvalErrorPrediction = 0.51326700 * 1206; time = 0.5364s; samplesPerSecond = 2248.3
|
||||
MPI Rank 1: 08/16/2016 10:08:45: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.92784049 * 1217; EvalClassificationError = 0.52670501 * 1217; time = 0.5780s; samplesPerSecond = 2105.4
|
||||
MPI Rank 1: 08/16/2016 10:08:45: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.88585235 * 1206; EvalClassificationError = 0.51326700 * 1206; time = 0.5364s; samplesPerSecond = 2248.3
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.06 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 1.10 seconds since last report (0.00 seconds on comm.); 4899 samples processed by 2 workers (2419 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 4.44k samplesPerSecond , throughputPerWorker = 2.22k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:46: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.94884216 * 1213; EvalErrorPrediction = 0.53833471 * 1213; time = 0.5669s; samplesPerSecond = 2139.7
|
||||
MPI Rank 1: 08/16/2016 10:08:46: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.83490224 * 1147; EvalErrorPrediction = 0.51525719 * 1147; time = 0.4897s; samplesPerSecond = 2342.2
|
||||
MPI Rank 1: 08/16/2016 10:08:46: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.94884216 * 1213; EvalClassificationError = 0.53833471 * 1213; time = 0.5669s; samplesPerSecond = 2139.7
|
||||
MPI Rank 1: 08/16/2016 10:08:46: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.83490224 * 1147; EvalClassificationError = 0.51525719 * 1147; time = 0.4897s; samplesPerSecond = 2342.2
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.06-seconds latency this time; accumulated time on sync point = 0.12 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 1.08 seconds since last report (0.00 seconds on comm.); 4829 samples processed by 2 workers (2359 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 4.48k samplesPerSecond , throughputPerWorker = 2.24k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:47: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 2.01499862 * 1212; EvalErrorPrediction = 0.55445545 * 1212; time = 0.5891s; samplesPerSecond = 2057.2
|
||||
MPI Rank 1: 08/16/2016 10:08:47: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.87405000 * 809; EvalErrorPrediction = 0.52286774 * 809; time = 0.3539s; samplesPerSecond = 2285.8
|
||||
MPI Rank 1: 08/16/2016 10:08:47: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 2.01499862 * 1212; EvalClassificationError = 0.55445545 * 1212; time = 0.5891s; samplesPerSecond = 2057.2
|
||||
MPI Rank 1: 08/16/2016 10:08:47: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.87405000 * 809; EvalClassificationError = 0.52286774 * 809; time = 0.3539s; samplesPerSecond = 2285.8
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.78-seconds latency this time; accumulated time on sync point = 0.90 seconds , average latency = 0.22 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 2.20 seconds since last report (1.05 seconds on comm.); 5833 samples processed by 2 workers (809 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 2.65k samplesPerSecond , throughputPerWorker = 1.33k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:49: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.87480299 * 20480; EvalErrorPrediction = 0.51845703 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-05; epochTime=5.50168s
|
||||
MPI Rank 1: 08/16/2016 10:08:49: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.87480299 * 20480; EvalClassificationError = 0.51845703 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-05; epochTime=5.50168s
|
||||
MPI Rank 1: 08/16/2016 10:08:49: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:49: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -105,7 +105,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -131,7 +131,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -213,7 +213,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -239,7 +239,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -327,7 +327,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -353,7 +353,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -453,7 +453,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -482,7 +482,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -505,7 +505,7 @@ MPI Rank 0: 08/16/2016 10:08:54: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 10:08:54: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:54: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 10:08:54: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 10:08:54: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -558,113 +558,113 @@ MPI Rank 0: 08/16/2016 10:08:57: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.68332137 * 192; EvalErrorPrediction = 0.98958333 * 192; time = 0.0211s; samplesPerSecond = 9080.6
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.42217834 * 192; EvalErrorPrediction = 0.89062500 * 192; time = 0.0200s; samplesPerSecond = 9581.8
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.78361173 * 192; EvalErrorPrediction = 0.93229167 * 192; time = 0.0200s; samplesPerSecond = 9583.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.28266373 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0200s; samplesPerSecond = 9584.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.48941255 * 192; EvalErrorPrediction = 0.89583333 * 192; time = 0.0200s; samplesPerSecond = 9589.5
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.62%]: CrossEntropyWithSoftmax = 4.39312485 * 192; EvalErrorPrediction = 0.94270833 * 192; time = 0.0200s; samplesPerSecond = 9583.7
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.26351131 * 192; EvalErrorPrediction = 0.98437500 * 192; time = 0.0198s; samplesPerSecond = 9681.3
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06459300 * 192; EvalErrorPrediction = 0.90625000 * 192; time = 0.0195s; samplesPerSecond = 9859.3
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97918086 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0195s; samplesPerSecond = 9859.3
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83987406 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79421365 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0195s; samplesPerSecond = 9857.3
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.76043875 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9855.3
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73154819 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9854.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.12%]: CrossEntropyWithSoftmax = 3.68435113 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89304289 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.76826980 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81256010 * 192; EvalErrorPrediction = 0.91145833 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.92133036 * 192; EvalErrorPrediction = 0.87500000 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56128851 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.67830123 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.36612487 * 192; EvalErrorPrediction = 0.76041667 * 192; time = 0.0195s; samplesPerSecond = 9843.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.62%]: CrossEntropyWithSoftmax = 3.41785793 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9858.8
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.37660990 * 192; EvalErrorPrediction = 0.79166667 * 192; time = 0.0195s; samplesPerSecond = 9842.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.68727319 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9836.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.64994161 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9868.4
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.37700933 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51711918 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47828208 * 192; EvalErrorPrediction = 0.80208333 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.54864536 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9843.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.12%]: CrossEntropyWithSoftmax = 3.43454656 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.36875092 * 192; EvalErrorPrediction = 0.79687500 * 192; time = 0.0195s; samplesPerSecond = 9856.3
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55401747 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0195s; samplesPerSecond = 9847.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.34809828 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9843.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.49450178 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.55445642 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9859.3
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.49293717 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.32494958 * 192; EvalErrorPrediction = 0.79687500 * 192; time = 0.0195s; samplesPerSecond = 9836.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.62%]: CrossEntropyWithSoftmax = 3.28851645 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9838.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.19411841 * 192; EvalErrorPrediction = 0.78125000 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.25028950 * 192; EvalErrorPrediction = 0.78125000 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.53445091 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9840.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00326214 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0195s; samplesPerSecond = 9853.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.10787619 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.05588285 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18197888 * 192; EvalErrorPrediction = 0.74479167 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.12%]: CrossEntropyWithSoftmax = 2.78138232 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9856.8
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17441293 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03537426 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.00595982 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0195s; samplesPerSecond = 9833.0
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79115323 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.04097805 * 192; EvalErrorPrediction = 0.69791667 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.15935323 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11947251 * 192; EvalErrorPrediction = 0.75000000 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.62%]: CrossEntropyWithSoftmax = 2.86232384 * 192; EvalErrorPrediction = 0.72395833 * 192; time = 0.0195s; samplesPerSecond = 9843.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.77742114 * 192; EvalErrorPrediction = 0.70312500 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00411374 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69740857 * 192; EvalErrorPrediction = 0.68750000 * 192; time = 0.0195s; samplesPerSecond = 9838.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64948203 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9830.0
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71417618 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9839.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.74281938 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9851.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.81346277 * 192; EvalErrorPrediction = 0.72916667 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79862588 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0195s; samplesPerSecond = 9858.3
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.76655026 * 192; EvalErrorPrediction = 0.71354167 * 192; time = 0.0195s; samplesPerSecond = 9857.3
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.75908943 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.53548199 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97589950 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.64996722 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9845.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52128374 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.64228785 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0195s; samplesPerSecond = 9855.3
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.62%]: CrossEntropyWithSoftmax = 2.57199182 * 192; EvalErrorPrediction = 0.67708333 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.58100902 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.48555431 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.75336137 * 192; EvalErrorPrediction = 0.67187500 * 192; time = 0.0195s; samplesPerSecond = 9844.6
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.49193178 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.46098647 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.74322101 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.55837089 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.12%]: CrossEntropyWithSoftmax = 2.98288510 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9851.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.36667287 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0195s; samplesPerSecond = 9849.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.25169614 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.31564120 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.70894816 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9847.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.44991146 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.51856232 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.38498228 * 192; EvalErrorPrediction = 0.61979167 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.62%]: CrossEntropyWithSoftmax = 2.38080698 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.34294967 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.30340167 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0195s; samplesPerSecond = 9854.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.08323277 * 192; EvalErrorPrediction = 0.52604167 * 192; time = 0.0195s; samplesPerSecond = 9844.1
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.43589953 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.53399486 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33995708 * 192; EvalErrorPrediction = 0.60416667 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.62970864 * 192; EvalErrorPrediction = 0.68229167 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.12%]: CrossEntropyWithSoftmax = 2.47993989 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9843.6
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30935254 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.22022265 * 192; EvalErrorPrediction = 0.59375000 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.28060247 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0195s; samplesPerSecond = 9838.6
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.13349744 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.29751600 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.29319143 * 192; EvalErrorPrediction = 0.60416667 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.14551002 * 192; EvalErrorPrediction = 0.55729167 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.62%]: CrossEntropyWithSoftmax = 2.26930674 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.17383355 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.43111882 * 192; EvalErrorPrediction = 0.67187500 * 192; time = 0.0195s; samplesPerSecond = 9854.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.18011227 * 192; EvalErrorPrediction = 0.59895833 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.21682707 * 192; EvalErrorPrediction = 0.56250000 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.09164s
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.68332137 * 192; EvalClassificationError = 0.98958333 * 192; time = 0.0211s; samplesPerSecond = 9080.6
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.42217834 * 192; EvalClassificationError = 0.89062500 * 192; time = 0.0200s; samplesPerSecond = 9581.8
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.78361173 * 192; EvalClassificationError = 0.93229167 * 192; time = 0.0200s; samplesPerSecond = 9583.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.28266373 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0200s; samplesPerSecond = 9584.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.48941255 * 192; EvalClassificationError = 0.89583333 * 192; time = 0.0200s; samplesPerSecond = 9589.5
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.62%]: CrossEntropyWithSoftmax = 4.39312485 * 192; EvalClassificationError = 0.94270833 * 192; time = 0.0200s; samplesPerSecond = 9583.7
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.26351131 * 192; EvalClassificationError = 0.98437500 * 192; time = 0.0198s; samplesPerSecond = 9681.3
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06459300 * 192; EvalClassificationError = 0.90625000 * 192; time = 0.0195s; samplesPerSecond = 9859.3
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97918086 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0195s; samplesPerSecond = 9859.3
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83987406 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79421365 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0195s; samplesPerSecond = 9857.3
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.76043875 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9855.3
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73154819 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9854.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.12%]: CrossEntropyWithSoftmax = 3.68435113 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89304289 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.76826980 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81256010 * 192; EvalClassificationError = 0.91145833 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.92133036 * 192; EvalClassificationError = 0.87500000 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56128851 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.67830123 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.36612487 * 192; EvalClassificationError = 0.76041667 * 192; time = 0.0195s; samplesPerSecond = 9843.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.62%]: CrossEntropyWithSoftmax = 3.41785793 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9858.8
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.37660990 * 192; EvalClassificationError = 0.79166667 * 192; time = 0.0195s; samplesPerSecond = 9842.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.68727319 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9836.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.64994161 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9868.4
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.37700933 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51711918 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47828208 * 192; EvalClassificationError = 0.80208333 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.54864536 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9843.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.12%]: CrossEntropyWithSoftmax = 3.43454656 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.36875092 * 192; EvalClassificationError = 0.79687500 * 192; time = 0.0195s; samplesPerSecond = 9856.3
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55401747 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0195s; samplesPerSecond = 9847.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.34809828 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9843.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.49450178 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.55445642 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9859.3
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.49293717 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.32494958 * 192; EvalClassificationError = 0.79687500 * 192; time = 0.0195s; samplesPerSecond = 9836.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.62%]: CrossEntropyWithSoftmax = 3.28851645 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9838.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.19411841 * 192; EvalClassificationError = 0.78125000 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.25028950 * 192; EvalClassificationError = 0.78125000 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.53445091 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9840.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00326214 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0195s; samplesPerSecond = 9853.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.10787619 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.05588285 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18197888 * 192; EvalClassificationError = 0.74479167 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.12%]: CrossEntropyWithSoftmax = 2.78138232 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9856.8
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17441293 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03537426 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.00595982 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0195s; samplesPerSecond = 9833.0
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79115323 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.04097805 * 192; EvalClassificationError = 0.69791667 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.15935323 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11947251 * 192; EvalClassificationError = 0.75000000 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.62%]: CrossEntropyWithSoftmax = 2.86232384 * 192; EvalClassificationError = 0.72395833 * 192; time = 0.0195s; samplesPerSecond = 9843.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.77742114 * 192; EvalClassificationError = 0.70312500 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00411374 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69740857 * 192; EvalClassificationError = 0.68750000 * 192; time = 0.0195s; samplesPerSecond = 9838.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64948203 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9830.0
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71417618 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9839.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.74281938 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9851.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.81346277 * 192; EvalClassificationError = 0.72916667 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79862588 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0195s; samplesPerSecond = 9858.3
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.76655026 * 192; EvalClassificationError = 0.71354167 * 192; time = 0.0195s; samplesPerSecond = 9857.3
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.75908943 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.53548199 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97589950 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.64996722 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9845.1
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52128374 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.64228785 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0195s; samplesPerSecond = 9855.3
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.62%]: CrossEntropyWithSoftmax = 2.57199182 * 192; EvalClassificationError = 0.67708333 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.58100902 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.48555431 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.75336137 * 192; EvalClassificationError = 0.67187500 * 192; time = 0.0195s; samplesPerSecond = 9844.6
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.49193178 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.46098647 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.74322101 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.55837089 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.12%]: CrossEntropyWithSoftmax = 2.98288510 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9851.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.36667287 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0195s; samplesPerSecond = 9849.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.25169614 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.31564120 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.70894816 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9847.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.44991146 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.51856232 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.38498228 * 192; EvalClassificationError = 0.61979167 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.62%]: CrossEntropyWithSoftmax = 2.38080698 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.34294967 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.30340167 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0195s; samplesPerSecond = 9854.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.08323277 * 192; EvalClassificationError = 0.52604167 * 192; time = 0.0195s; samplesPerSecond = 9844.1
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.43589953 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.53399486 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33995708 * 192; EvalClassificationError = 0.60416667 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.62970864 * 192; EvalClassificationError = 0.68229167 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.12%]: CrossEntropyWithSoftmax = 2.47993989 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9843.6
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30935254 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.22022265 * 192; EvalClassificationError = 0.59375000 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.28060247 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0195s; samplesPerSecond = 9838.6
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.13349744 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.29751600 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.29319143 * 192; EvalClassificationError = 0.60416667 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.14551002 * 192; EvalClassificationError = 0.55729167 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.62%]: CrossEntropyWithSoftmax = 2.26930674 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.17383355 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.43111882 * 192; EvalClassificationError = 0.67187500 * 192; time = 0.0195s; samplesPerSecond = 9854.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.18011227 * 192; EvalClassificationError = 0.59895833 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.21682707 * 192; EvalClassificationError = 0.56250000 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.09164s
|
||||
MPI Rank 0: 08/16/2016 10:08:59: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBM@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -672,46 +672,46 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.20431953 * 508; EvalErrorPrediction = 0.59842520 * 508; time = 0.0245s; samplesPerSecond = 20758.4
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.19237836 * 492; EvalErrorPrediction = 0.57723577 * 492; time = 0.0207s; samplesPerSecond = 23745.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.17526222 * 488; EvalErrorPrediction = 0.59836066 * 488; time = 0.0259s; samplesPerSecond = 18826.4
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.23483521 * 527; EvalErrorPrediction = 0.59582543 * 527; time = 0.0385s; samplesPerSecond = 13687.6
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.20431953 * 508; EvalClassificationError = 0.59842520 * 508; time = 0.0245s; samplesPerSecond = 20758.4
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.19237836 * 492; EvalClassificationError = 0.57723577 * 492; time = 0.0207s; samplesPerSecond = 23745.2
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.17526222 * 488; EvalClassificationError = 0.59836066 * 488; time = 0.0259s; samplesPerSecond = 18826.4
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.23483521 * 527; EvalClassificationError = 0.59582543 * 527; time = 0.0385s; samplesPerSecond = 13687.6
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.04-seconds latency this time; accumulated time on sync point = 0.04 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.18 seconds since last report (0.01 seconds on comm.); 4289 samples processed by 2 workers (2163 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 24.12k samplesPerSecond , throughputPerWorker = 12.06k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.00528284 * 473; EvalErrorPrediction = 0.53911205 * 473; time = 0.0836s; samplesPerSecond = 5659.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.08558038 * 511; EvalErrorPrediction = 0.54990215 * 511; time = 0.0286s; samplesPerSecond = 17845.9
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.08506441 * 506; EvalErrorPrediction = 0.54940711 * 506; time = 0.0272s; samplesPerSecond = 18597.5
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.12168501 * 513; EvalErrorPrediction = 0.52241715 * 513; time = 0.0287s; samplesPerSecond = 17902.0
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.00528284 * 473; EvalClassificationError = 0.53911205 * 473; time = 0.0836s; samplesPerSecond = 5659.7
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.08558038 * 511; EvalClassificationError = 0.54990215 * 511; time = 0.0286s; samplesPerSecond = 17845.9
|
||||
MPI Rank 0: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.08506441 * 506; EvalClassificationError = 0.54940711 * 506; time = 0.0272s; samplesPerSecond = 18597.5
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.12168501 * 513; EvalClassificationError = 0.52241715 * 513; time = 0.0287s; samplesPerSecond = 17902.0
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.08 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.17 seconds since last report (0.01 seconds on comm.); 4253 samples processed by 2 workers (2180 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 25.18k samplesPerSecond , throughputPerWorker = 12.59k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.08058681 * 489; EvalErrorPrediction = 0.56646217 * 489; time = 0.0753s; samplesPerSecond = 6497.0
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07411044 * 494; EvalErrorPrediction = 0.55060729 * 494; time = 0.0290s; samplesPerSecond = 17021.6
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 2.12310728 * 499; EvalErrorPrediction = 0.58316633 * 499; time = 0.0297s; samplesPerSecond = 16805.9
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.06918046 * 490; EvalErrorPrediction = 0.57142857 * 490; time = 0.0268s; samplesPerSecond = 18265.2
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.08058681 * 489; EvalClassificationError = 0.56646217 * 489; time = 0.0753s; samplesPerSecond = 6497.0
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07411044 * 494; EvalClassificationError = 0.55060729 * 494; time = 0.0290s; samplesPerSecond = 17021.6
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 2.12310728 * 499; EvalClassificationError = 0.58316633 * 499; time = 0.0297s; samplesPerSecond = 16805.9
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.06918046 * 490; EvalClassificationError = 0.57142857 * 490; time = 0.0268s; samplesPerSecond = 18265.2
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.10 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.17 seconds since last report (0.01 seconds on comm.); 4246 samples processed by 2 workers (2144 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 25.14k samplesPerSecond , throughputPerWorker = 12.57k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 1.93293190 * 497; EvalErrorPrediction = 0.50905433 * 497; time = 0.0737s; samplesPerSecond = 6740.7
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.13718669 * 492; EvalErrorPrediction = 0.59552846 * 492; time = 0.0287s; samplesPerSecond = 17130.3
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 1.91004653 * 508; EvalErrorPrediction = 0.54527559 * 508; time = 0.0272s; samplesPerSecond = 18644.3
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.97341931 * 503; EvalErrorPrediction = 0.52286282 * 503; time = 0.0256s; samplesPerSecond = 19615.5
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.07837958 * 470; EvalErrorPrediction = 0.57021277 * 470; time = 0.0152s; samplesPerSecond = 30933.3
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.91466321 * 494; EvalErrorPrediction = 0.54858300 * 494; time = 0.0156s; samplesPerSecond = 31693.1
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 2.00598279 * 503; EvalErrorPrediction = 0.52683897 * 503; time = 0.0183s; samplesPerSecond = 27514.9
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96239369 * 487; EvalErrorPrediction = 0.54004107 * 487; time = 0.0154s; samplesPerSecond = 31537.4
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.96513086 * 516; EvalErrorPrediction = 0.51744186 * 516; time = 0.0158s; samplesPerSecond = 32687.2
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 1.97088895 * 494; EvalErrorPrediction = 0.55668016 * 494; time = 0.0156s; samplesPerSecond = 31632.2
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 2.01240053 * 510; EvalErrorPrediction = 0.56470588 * 510; time = 0.0157s; samplesPerSecond = 32486.1
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 2.02387287 * 497; EvalErrorPrediction = 0.54929577 * 497; time = 0.0155s; samplesPerSecond = 32021.1
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.90663199 * 490; EvalErrorPrediction = 0.55306122 * 490; time = 0.0155s; samplesPerSecond = 31670.1
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.94815649 * 482; EvalErrorPrediction = 0.52697095 * 482; time = 0.0154s; samplesPerSecond = 31325.1
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.97332721 * 342; EvalErrorPrediction = 0.50877193 * 342; time = 0.0104s; samplesPerSecond = 32758.6
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 1.93293190 * 497; EvalClassificationError = 0.50905433 * 497; time = 0.0737s; samplesPerSecond = 6740.7
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.13718669 * 492; EvalClassificationError = 0.59552846 * 492; time = 0.0287s; samplesPerSecond = 17130.3
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 1.91004653 * 508; EvalClassificationError = 0.54527559 * 508; time = 0.0272s; samplesPerSecond = 18644.3
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.97341931 * 503; EvalClassificationError = 0.52286282 * 503; time = 0.0256s; samplesPerSecond = 19615.5
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.07837958 * 470; EvalClassificationError = 0.57021277 * 470; time = 0.0152s; samplesPerSecond = 30933.3
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.91466321 * 494; EvalClassificationError = 0.54858300 * 494; time = 0.0156s; samplesPerSecond = 31693.1
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 2.00598279 * 503; EvalClassificationError = 0.52683897 * 503; time = 0.0183s; samplesPerSecond = 27514.9
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96239369 * 487; EvalClassificationError = 0.54004107 * 487; time = 0.0154s; samplesPerSecond = 31537.4
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.96513086 * 516; EvalClassificationError = 0.51744186 * 516; time = 0.0158s; samplesPerSecond = 32687.2
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 1.97088895 * 494; EvalClassificationError = 0.55668016 * 494; time = 0.0156s; samplesPerSecond = 31632.2
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 2.01240053 * 510; EvalClassificationError = 0.56470588 * 510; time = 0.0157s; samplesPerSecond = 32486.1
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 2.02387287 * 497; EvalClassificationError = 0.54929577 * 497; time = 0.0155s; samplesPerSecond = 32021.1
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.90663199 * 490; EvalClassificationError = 0.55306122 * 490; time = 0.0155s; samplesPerSecond = 31670.1
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.94815649 * 482; EvalClassificationError = 0.52697095 * 482; time = 0.0154s; samplesPerSecond = 31325.1
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.97332721 * 342; EvalClassificationError = 0.50877193 * 342; time = 0.0104s; samplesPerSecond = 32758.6
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.10 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.27 seconds since last report (0.01 seconds on comm.); 7692 samples processed by 2 workers (6788 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 28.58k samplesPerSecond , throughputPerWorker = 14.29k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.05172118 * 20480; EvalErrorPrediction = 0.55805664 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.785025s
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.05172118 * 20480; EvalClassificationError = 0.55805664 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.785025s
|
||||
MPI Rank 0: 08/16/2016 10:09:00: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBM@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -719,26 +719,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.96185095 * 1942; EvalErrorPrediction = 0.53347065 * 1942; time = 0.0736s; samplesPerSecond = 26393.8
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.96185095 * 1942; EvalClassificationError = 0.53347065 * 1942; time = 0.0736s; samplesPerSecond = 26393.8
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.04-seconds latency this time; accumulated time on sync point = 0.04 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.13 seconds since last report (0.01 seconds on comm.); 4885 samples processed by 2 workers (2592 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 36.46k samplesPerSecond , throughputPerWorker = 18.23k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94171416 * 1909; EvalErrorPrediction = 0.55055003 * 1909; time = 0.1085s; samplesPerSecond = 17588.6
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94171416 * 1909; EvalClassificationError = 0.55055003 * 1909; time = 0.1085s; samplesPerSecond = 17588.6
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.04 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.12 seconds since last report (0.00 seconds on comm.); 4826 samples processed by 2 workers (2577 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 39.11k samplesPerSecond , throughputPerWorker = 19.55k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.98907844 * 1987; EvalErrorPrediction = 0.55158530 * 1987; time = 0.1006s; samplesPerSecond = 19757.0
|
||||
MPI Rank 0: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.98907844 * 1987; EvalClassificationError = 0.55158530 * 1987; time = 0.1006s; samplesPerSecond = 19757.0
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.06 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.12 seconds since last report (0.01 seconds on comm.); 4903 samples processed by 2 workers (2577 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 40.04k samplesPerSecond , throughputPerWorker = 20.02k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.94333646 * 1908; EvalErrorPrediction = 0.54926625 * 1908; time = 0.0927s; samplesPerSecond = 20579.0
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.97418902 * 1905; EvalErrorPrediction = 0.55223097 * 1905; time = 0.0634s; samplesPerSecond = 30038.3
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.96248532 * 1913; EvalErrorPrediction = 0.54887611 * 1913; time = 0.0447s; samplesPerSecond = 42749.6
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.97409307 * 1225; EvalErrorPrediction = 0.54367347 * 1225; time = 0.0288s; samplesPerSecond = 42593.9
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.94333646 * 1908; EvalClassificationError = 0.54926625 * 1908; time = 0.0927s; samplesPerSecond = 20579.0
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.97418902 * 1905; EvalClassificationError = 0.55223097 * 1905; time = 0.0634s; samplesPerSecond = 30038.3
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.96248532 * 1913; EvalClassificationError = 0.54887611 * 1913; time = 0.0447s; samplesPerSecond = 42749.6
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.97409307 * 1225; EvalClassificationError = 0.54367347 * 1225; time = 0.0288s; samplesPerSecond = 42593.9
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.06 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.16 seconds since last report (0.01 seconds on comm.); 5866 samples processed by 2 workers (5043 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 37.73k samplesPerSecond , throughputPerWorker = 18.86k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.95703393 * 20480; EvalErrorPrediction = 0.54541016 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.535515s
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.95703393 * 20480; EvalClassificationError = 0.54541016 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.535515s
|
||||
MPI Rank 0: 08/16/2016 10:09:01: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBM@release_gpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -746,26 +746,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.88963761 * 1923; EvalErrorPrediction = 0.52366095 * 1923; time = 0.0808s; samplesPerSecond = 23804.5
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.88963761 * 1923; EvalClassificationError = 0.52366095 * 1923; time = 0.0808s; samplesPerSecond = 23804.5
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.13 seconds since last report (0.01 seconds on comm.); 4901 samples processed by 2 workers (2550 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 38.02k samplesPerSecond , throughputPerWorker = 19.01k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.89669303 * 1870; EvalErrorPrediction = 0.51871658 * 1870; time = 0.0946s; samplesPerSecond = 19777.3
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.89669303 * 1870; EvalClassificationError = 0.51871658 * 1870; time = 0.0946s; samplesPerSecond = 19777.3
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.05 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.12 seconds since last report (0.01 seconds on comm.); 4836 samples processed by 2 workers (2519 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 40.33k samplesPerSecond , throughputPerWorker = 20.17k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.91111689 * 1942; EvalErrorPrediction = 0.54119464 * 1942; time = 0.1021s; samplesPerSecond = 19015.2
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.91111689 * 1942; EvalClassificationError = 0.54119464 * 1942; time = 0.1021s; samplesPerSecond = 19015.2
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.01-seconds latency this time; accumulated time on sync point = 0.06 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.12 seconds since last report (0.01 seconds on comm.); 4952 samples processed by 2 workers (2551 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 40.08k samplesPerSecond , throughputPerWorker = 20.04k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.88475158 * 1885; EvalErrorPrediction = 0.51458886 * 1885; time = 0.0900s; samplesPerSecond = 20941.2
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.89480846 * 1870; EvalErrorPrediction = 0.51497326 * 1870; time = 0.0626s; samplesPerSecond = 29858.4
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.89539137 * 1873; EvalErrorPrediction = 0.52108916 * 1873; time = 0.0432s; samplesPerSecond = 43385.6
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.93128839 * 1231; EvalErrorPrediction = 0.52721365 * 1231; time = 0.0285s; samplesPerSecond = 43159.7
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.88475158 * 1885; EvalClassificationError = 0.51458886 * 1885; time = 0.0900s; samplesPerSecond = 20941.2
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.89480846 * 1870; EvalClassificationError = 0.51497326 * 1870; time = 0.0626s; samplesPerSecond = 29858.4
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.89539137 * 1873; EvalClassificationError = 0.52108916 * 1873; time = 0.0432s; samplesPerSecond = 43385.6
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.93128839 * 1231; EvalClassificationError = 0.52721365 * 1231; time = 0.0285s; samplesPerSecond = 43159.7
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.06 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.14 seconds since last report (0.00 seconds on comm.); 5791 samples processed by 2 workers (4974 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 40.15k samplesPerSecond , throughputPerWorker = 20.08k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.90306770 * 20480; EvalErrorPrediction = 0.52641602 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.516692s
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.90306770 * 20480; EvalClassificationError = 0.52641602 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.516692s
|
||||
MPI Rank 0: 08/16/2016 10:09:01: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBM@release_gpu/models/cntkSpeech.dnn.4'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Starting Epoch 5: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -773,26 +773,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 4: frames [81920..102400] (first utterance at frame 81920), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.93021270 * 1863; EvalErrorPrediction = 0.51851852 * 1863; time = 0.0761s; samplesPerSecond = 24471.9
|
||||
MPI Rank 0: 08/16/2016 10:09:01: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.93021270 * 1863; EvalClassificationError = 0.51851852 * 1863; time = 0.0761s; samplesPerSecond = 24471.9
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.03 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.14 seconds since last report (0.02 seconds on comm.); 4919 samples processed by 2 workers (2493 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 35.45k samplesPerSecond , throughputPerWorker = 17.72k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.87273976 * 1855; EvalErrorPrediction = 0.52129380 * 1855; time = 0.1110s; samplesPerSecond = 16709.3
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.87273976 * 1855; EvalClassificationError = 0.52129380 * 1855; time = 0.1110s; samplesPerSecond = 16709.3
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.06 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.13 seconds since last report (0.02 seconds on comm.); 4899 samples processed by 2 workers (2480 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 37.61k samplesPerSecond , throughputPerWorker = 18.80k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.87195439 * 1866; EvalErrorPrediction = 0.52304394 * 1866; time = 0.1034s; samplesPerSecond = 18040.3
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.87195439 * 1866; EvalClassificationError = 0.52304394 * 1866; time = 0.1034s; samplesPerSecond = 18040.3
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.01-seconds latency this time; accumulated time on sync point = 0.07 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.13 seconds since last report (0.01 seconds on comm.); 4829 samples processed by 2 workers (2470 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 37.39k samplesPerSecond , throughputPerWorker = 18.70k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.88262131 * 1859; EvalErrorPrediction = 0.52017214 * 1859; time = 0.1034s; samplesPerSecond = 17972.8
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.82076948 * 1925; EvalErrorPrediction = 0.49714286 * 1925; time = 0.0628s; samplesPerSecond = 30643.6
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.84625728 * 1860; EvalErrorPrediction = 0.50967742 * 1860; time = 0.0439s; samplesPerSecond = 42354.5
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.86158884 * 1239; EvalErrorPrediction = 0.51412429 * 1239; time = 0.0287s; samplesPerSecond = 43128.7
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.88262131 * 1859; EvalClassificationError = 0.52017214 * 1859; time = 0.1034s; samplesPerSecond = 17972.8
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.82076948 * 1925; EvalClassificationError = 0.49714286 * 1925; time = 0.0628s; samplesPerSecond = 30643.6
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.84625728 * 1860; EvalClassificationError = 0.50967742 * 1860; time = 0.0439s; samplesPerSecond = 42354.5
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.86158884 * 1239; EvalClassificationError = 0.51412429 * 1239; time = 0.0287s; samplesPerSecond = 43128.7
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.07 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.15 seconds since last report (0.01 seconds on comm.); 5833 samples processed by 2 workers (5024 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 37.93k samplesPerSecond , throughputPerWorker = 18.96k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.88963745 * 20480; EvalErrorPrediction = 0.51865234 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-05; epochTime=0.552101s
|
||||
MPI Rank 0: 08/16/2016 10:09:02: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.88963745 * 20480; EvalClassificationError = 0.51865234 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-05; epochTime=0.552101s
|
||||
MPI Rank 0: 08/16/2016 10:09:02: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBM@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 10:09:02: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -845,7 +845,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -871,7 +871,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -953,7 +953,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -979,7 +979,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1067,7 +1067,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1093,7 +1093,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1193,7 +1193,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1222,7 +1222,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1245,7 +1245,7 @@ MPI Rank 1: 08/16/2016 10:08:55: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 10:08:55: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:55: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 10:08:55: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 10:08:55: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1298,237 +1298,237 @@ MPI Rank 1: 08/16/2016 10:08:57: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.68332137 * 192; EvalErrorPrediction = 0.98958333 * 192; time = 0.0210s; samplesPerSecond = 9126.8
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.42217834 * 192; EvalErrorPrediction = 0.89062500 * 192; time = 0.0200s; samplesPerSecond = 9580.4
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.78361173 * 192; EvalErrorPrediction = 0.93229167 * 192; time = 0.0200s; samplesPerSecond = 9585.1
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.28266373 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0200s; samplesPerSecond = 9584.2
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.48941255 * 192; EvalErrorPrediction = 0.89583333 * 192; time = 0.0200s; samplesPerSecond = 9588.0
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.62%]: CrossEntropyWithSoftmax = 4.39312485 * 192; EvalErrorPrediction = 0.94270833 * 192; time = 0.0200s; samplesPerSecond = 9584.2
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.26351131 * 192; EvalErrorPrediction = 0.98437500 * 192; time = 0.0192s; samplesPerSecond = 10025.1
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06459300 * 192; EvalErrorPrediction = 0.90625000 * 192; time = 0.0195s; samplesPerSecond = 9856.8
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97918086 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0195s; samplesPerSecond = 9853.7
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83987406 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79421365 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.76043875 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9858.3
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73154819 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9855.3
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.12%]: CrossEntropyWithSoftmax = 3.68435113 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89304289 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0195s; samplesPerSecond = 9849.2
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.76826980 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81256010 * 192; EvalErrorPrediction = 0.91145833 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.92133036 * 192; EvalErrorPrediction = 0.87500000 * 192; time = 0.0195s; samplesPerSecond = 9848.7
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56128851 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0195s; samplesPerSecond = 9856.8
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.67830123 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9845.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.36612487 * 192; EvalErrorPrediction = 0.76041667 * 192; time = 0.0195s; samplesPerSecond = 9848.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.62%]: CrossEntropyWithSoftmax = 3.41785793 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.37660990 * 192; EvalErrorPrediction = 0.79166667 * 192; time = 0.0195s; samplesPerSecond = 9842.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.68727319 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9844.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.64994161 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.37700933 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51711918 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47828208 * 192; EvalErrorPrediction = 0.80208333 * 192; time = 0.0195s; samplesPerSecond = 9854.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.54864536 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9838.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.12%]: CrossEntropyWithSoftmax = 3.43454656 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9851.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.36875092 * 192; EvalErrorPrediction = 0.79687500 * 192; time = 0.0195s; samplesPerSecond = 9858.3
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55401747 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.34809828 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9838.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.49450178 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9849.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.55445642 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.49293717 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9844.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.32494958 * 192; EvalErrorPrediction = 0.79687500 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.62%]: CrossEntropyWithSoftmax = 3.28851645 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9837.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.19411841 * 192; EvalErrorPrediction = 0.78125000 * 192; time = 0.0195s; samplesPerSecond = 9851.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.25028950 * 192; EvalErrorPrediction = 0.78125000 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.53445091 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9849.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00326214 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.10787619 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.05588285 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18197888 * 192; EvalErrorPrediction = 0.74479167 * 192; time = 0.0195s; samplesPerSecond = 9841.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.12%]: CrossEntropyWithSoftmax = 2.78138232 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17441293 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0195s; samplesPerSecond = 9843.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03537426 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0195s; samplesPerSecond = 9853.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.00595982 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0195s; samplesPerSecond = 9840.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79115323 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9842.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.04097805 * 192; EvalErrorPrediction = 0.69791667 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.15935323 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11947251 * 192; EvalErrorPrediction = 0.75000000 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.62%]: CrossEntropyWithSoftmax = 2.86232384 * 192; EvalErrorPrediction = 0.72395833 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.77742114 * 192; EvalErrorPrediction = 0.70312500 * 192; time = 0.0195s; samplesPerSecond = 9841.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00411374 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0195s; samplesPerSecond = 9848.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69740857 * 192; EvalErrorPrediction = 0.68750000 * 192; time = 0.0195s; samplesPerSecond = 9838.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64948203 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9835.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71417618 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9837.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.74281938 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9845.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.81346277 * 192; EvalErrorPrediction = 0.72916667 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79862588 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0195s; samplesPerSecond = 9857.3
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.76655026 * 192; EvalErrorPrediction = 0.71354167 * 192; time = 0.0195s; samplesPerSecond = 9856.3
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.75908943 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.53548199 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97589950 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0195s; samplesPerSecond = 9848.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.64996722 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52128374 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.64228785 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0195s; samplesPerSecond = 9853.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.62%]: CrossEntropyWithSoftmax = 2.57199182 * 192; EvalErrorPrediction = 0.67708333 * 192; time = 0.0195s; samplesPerSecond = 9848.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.58100902 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.48555431 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9855.3
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.75336137 * 192; EvalErrorPrediction = 0.67187500 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.49193178 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9847.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.46098647 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.74322101 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.55837089 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9848.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.12%]: CrossEntropyWithSoftmax = 2.98288510 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.36667287 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.25169614 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.31564120 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.70894816 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9848.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.44991146 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9845.1
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.51856232 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.38498228 * 192; EvalErrorPrediction = 0.61979167 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.62%]: CrossEntropyWithSoftmax = 2.38080698 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.34294967 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.30340167 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.08323277 * 192; EvalErrorPrediction = 0.52604167 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.43589953 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.53399486 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33995708 * 192; EvalErrorPrediction = 0.60416667 * 192; time = 0.0195s; samplesPerSecond = 9847.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.62970864 * 192; EvalErrorPrediction = 0.68229167 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.12%]: CrossEntropyWithSoftmax = 2.47993989 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30935254 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.22022265 * 192; EvalErrorPrediction = 0.59375000 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.28060247 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0195s; samplesPerSecond = 9835.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.13349744 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.29751600 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9855.3
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.29319143 * 192; EvalErrorPrediction = 0.60416667 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.14551002 * 192; EvalErrorPrediction = 0.55729167 * 192; time = 0.0195s; samplesPerSecond = 9844.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.62%]: CrossEntropyWithSoftmax = 2.26930674 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0195s; samplesPerSecond = 9856.3
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.17383355 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.43111882 * 192; EvalErrorPrediction = 0.67187500 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.18011227 * 192; EvalErrorPrediction = 0.59895833 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.21682707 * 192; EvalErrorPrediction = 0.56250000 * 192; time = 0.0195s; samplesPerSecond = 9843.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.0909s
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.68332137 * 192; EvalClassificationError = 0.98958333 * 192; time = 0.0210s; samplesPerSecond = 9126.8
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.42217834 * 192; EvalClassificationError = 0.89062500 * 192; time = 0.0200s; samplesPerSecond = 9580.4
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.78361173 * 192; EvalClassificationError = 0.93229167 * 192; time = 0.0200s; samplesPerSecond = 9585.1
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.28266373 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0200s; samplesPerSecond = 9584.2
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.48941255 * 192; EvalClassificationError = 0.89583333 * 192; time = 0.0200s; samplesPerSecond = 9588.0
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.62%]: CrossEntropyWithSoftmax = 4.39312485 * 192; EvalClassificationError = 0.94270833 * 192; time = 0.0200s; samplesPerSecond = 9584.2
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.26351131 * 192; EvalClassificationError = 0.98437500 * 192; time = 0.0192s; samplesPerSecond = 10025.1
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06459300 * 192; EvalClassificationError = 0.90625000 * 192; time = 0.0195s; samplesPerSecond = 9856.8
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97918086 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0195s; samplesPerSecond = 9853.7
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83987406 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79421365 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.76043875 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9858.3
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73154819 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9855.3
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.12%]: CrossEntropyWithSoftmax = 3.68435113 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89304289 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0195s; samplesPerSecond = 9849.2
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.76826980 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81256010 * 192; EvalClassificationError = 0.91145833 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.92133036 * 192; EvalClassificationError = 0.87500000 * 192; time = 0.0195s; samplesPerSecond = 9848.7
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56128851 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0195s; samplesPerSecond = 9856.8
|
||||
MPI Rank 1: 08/16/2016 10:08:57: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.67830123 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0195s; samplesPerSecond = 9845.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.36612487 * 192; EvalClassificationError = 0.76041667 * 192; time = 0.0195s; samplesPerSecond = 9848.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.62%]: CrossEntropyWithSoftmax = 3.41785793 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.37660990 * 192; EvalClassificationError = 0.79166667 * 192; time = 0.0195s; samplesPerSecond = 9842.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.68727319 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9844.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.64994161 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.37700933 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51711918 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9851.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47828208 * 192; EvalClassificationError = 0.80208333 * 192; time = 0.0195s; samplesPerSecond = 9854.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.54864536 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9838.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.12%]: CrossEntropyWithSoftmax = 3.43454656 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9851.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.36875092 * 192; EvalClassificationError = 0.79687500 * 192; time = 0.0195s; samplesPerSecond = 9858.3
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55401747 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.34809828 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9838.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.49450178 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9849.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.55445642 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.49293717 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0195s; samplesPerSecond = 9844.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.32494958 * 192; EvalClassificationError = 0.79687500 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.62%]: CrossEntropyWithSoftmax = 3.28851645 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9837.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.19411841 * 192; EvalClassificationError = 0.78125000 * 192; time = 0.0195s; samplesPerSecond = 9851.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.25028950 * 192; EvalClassificationError = 0.78125000 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.53445091 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0195s; samplesPerSecond = 9849.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00326214 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.10787619 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0195s; samplesPerSecond = 9847.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.05588285 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9850.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18197888 * 192; EvalClassificationError = 0.74479167 * 192; time = 0.0195s; samplesPerSecond = 9841.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.12%]: CrossEntropyWithSoftmax = 2.78138232 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17441293 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0195s; samplesPerSecond = 9843.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03537426 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0195s; samplesPerSecond = 9853.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.00595982 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0195s; samplesPerSecond = 9840.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79115323 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9842.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.04097805 * 192; EvalClassificationError = 0.69791667 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.15935323 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11947251 * 192; EvalClassificationError = 0.75000000 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.62%]: CrossEntropyWithSoftmax = 2.86232384 * 192; EvalClassificationError = 0.72395833 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.77742114 * 192; EvalClassificationError = 0.70312500 * 192; time = 0.0195s; samplesPerSecond = 9841.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00411374 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0195s; samplesPerSecond = 9848.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69740857 * 192; EvalClassificationError = 0.68750000 * 192; time = 0.0195s; samplesPerSecond = 9838.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64948203 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9835.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71417618 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9837.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.74281938 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9845.1
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.81346277 * 192; EvalClassificationError = 0.72916667 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79862588 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0195s; samplesPerSecond = 9857.3
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.76655026 * 192; EvalClassificationError = 0.71354167 * 192; time = 0.0195s; samplesPerSecond = 9856.3
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.75908943 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.53548199 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97589950 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0195s; samplesPerSecond = 9848.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.64996722 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52128374 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.64228785 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0195s; samplesPerSecond = 9853.7
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.62%]: CrossEntropyWithSoftmax = 2.57199182 * 192; EvalClassificationError = 0.67708333 * 192; time = 0.0195s; samplesPerSecond = 9848.2
|
||||
MPI Rank 1: 08/16/2016 10:08:58: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.58100902 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.48555431 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9855.3
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.75336137 * 192; EvalClassificationError = 0.67187500 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.49193178 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9847.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.46098647 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.74322101 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9852.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.55837089 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9848.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.12%]: CrossEntropyWithSoftmax = 2.98288510 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.36667287 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.25169614 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.31564120 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0195s; samplesPerSecond = 9845.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.70894816 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0195s; samplesPerSecond = 9848.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.44991146 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0195s; samplesPerSecond = 9845.1
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.51856232 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.38498228 * 192; EvalClassificationError = 0.61979167 * 192; time = 0.0195s; samplesPerSecond = 9854.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.62%]: CrossEntropyWithSoftmax = 2.38080698 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.34294967 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9855.8
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.30340167 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.08323277 * 192; EvalClassificationError = 0.52604167 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.43589953 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.53399486 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33995708 * 192; EvalClassificationError = 0.60416667 * 192; time = 0.0195s; samplesPerSecond = 9847.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.62970864 * 192; EvalClassificationError = 0.68229167 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.12%]: CrossEntropyWithSoftmax = 2.47993989 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0195s; samplesPerSecond = 9850.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30935254 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.22022265 * 192; EvalClassificationError = 0.59375000 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.28060247 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0195s; samplesPerSecond = 9835.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.13349744 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0195s; samplesPerSecond = 9846.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.29751600 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0195s; samplesPerSecond = 9855.3
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.29319143 * 192; EvalClassificationError = 0.60416667 * 192; time = 0.0195s; samplesPerSecond = 9853.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.14551002 * 192; EvalClassificationError = 0.55729167 * 192; time = 0.0195s; samplesPerSecond = 9844.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.62%]: CrossEntropyWithSoftmax = 2.26930674 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0195s; samplesPerSecond = 9856.3
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.17383355 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0195s; samplesPerSecond = 9846.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.43111882 * 192; EvalClassificationError = 0.67187500 * 192; time = 0.0195s; samplesPerSecond = 9852.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.18011227 * 192; EvalClassificationError = 0.59895833 * 192; time = 0.0195s; samplesPerSecond = 9849.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.21682707 * 192; EvalClassificationError = 0.56250000 * 192; time = 0.0195s; samplesPerSecond = 9843.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.0909s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.19272896 * 260; EvalErrorPrediction = 0.61538462 * 260; time = 0.0310s; samplesPerSecond = 8376.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.34922865 * 276; EvalErrorPrediction = 0.70289855 * 276; time = 0.0324s; samplesPerSecond = 8524.0
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.24157888 * 280; EvalErrorPrediction = 0.67500000 * 280; time = 0.0197s; samplesPerSecond = 14247.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.20817680 * 241; EvalErrorPrediction = 0.62655602 * 241; time = 0.0157s; samplesPerSecond = 15326.9
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.04015325 * 295; EvalErrorPrediction = 0.52881356 * 295; time = 0.0253s; samplesPerSecond = 11675.3
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.40322904 * 257; EvalErrorPrediction = 0.63035019 * 257; time = 0.0108s; samplesPerSecond = 23737.0
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.04484358 * 262; EvalErrorPrediction = 0.56488550 * 262; time = 0.0109s; samplesPerSecond = 23999.3
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.19272896 * 260; EvalClassificationError = 0.61538462 * 260; time = 0.0310s; samplesPerSecond = 8376.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.34922865 * 276; EvalClassificationError = 0.70289855 * 276; time = 0.0324s; samplesPerSecond = 8524.0
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.24157888 * 280; EvalClassificationError = 0.67500000 * 280; time = 0.0197s; samplesPerSecond = 14247.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.20817680 * 241; EvalClassificationError = 0.62655602 * 241; time = 0.0157s; samplesPerSecond = 15326.9
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.04015325 * 295; EvalClassificationError = 0.52881356 * 295; time = 0.0253s; samplesPerSecond = 11675.3
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.40322904 * 257; EvalClassificationError = 0.63035019 * 257; time = 0.0108s; samplesPerSecond = 23737.0
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.04484358 * 262; EvalClassificationError = 0.56488550 * 262; time = 0.0109s; samplesPerSecond = 23999.3
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.18 seconds since last report (0.00 seconds on comm.); 4289 samples processed by 2 workers (2126 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 24.10k samplesPerSecond , throughputPerWorker = 12.05k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.03646309 * 255; EvalErrorPrediction = 0.53333333 * 255; time = 0.0295s; samplesPerSecond = 8644.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.07321525 * 279; EvalErrorPrediction = 0.58781362 * 279; time = 0.0243s; samplesPerSecond = 11486.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.06339373 * 274; EvalErrorPrediction = 0.52554745 * 274; time = 0.0234s; samplesPerSecond = 11709.9
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 1.95530128 * 269; EvalErrorPrediction = 0.53531599 * 269; time = 0.0248s; samplesPerSecond = 10841.5
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.03493597 * 278; EvalErrorPrediction = 0.61151079 * 278; time = 0.0226s; samplesPerSecond = 12314.5
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 2.13309941 * 271; EvalErrorPrediction = 0.54612546 * 271; time = 0.0260s; samplesPerSecond = 10429.9
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.19775096 * 276; EvalErrorPrediction = 0.58333333 * 276; time = 0.0111s; samplesPerSecond = 24950.3
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 2.14623378 * 260; EvalErrorPrediction = 0.58076923 * 260; time = 0.0109s; samplesPerSecond = 23903.6
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.03646309 * 255; EvalClassificationError = 0.53333333 * 255; time = 0.0295s; samplesPerSecond = 8644.7
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.07321525 * 279; EvalClassificationError = 0.58781362 * 279; time = 0.0243s; samplesPerSecond = 11486.2
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.06339373 * 274; EvalClassificationError = 0.52554745 * 274; time = 0.0234s; samplesPerSecond = 11709.9
|
||||
MPI Rank 1: 08/16/2016 10:08:59: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 1.95530128 * 269; EvalClassificationError = 0.53531599 * 269; time = 0.0248s; samplesPerSecond = 10841.5
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.03493597 * 278; EvalClassificationError = 0.61151079 * 278; time = 0.0226s; samplesPerSecond = 12314.5
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 2.13309941 * 271; EvalClassificationError = 0.54612546 * 271; time = 0.0260s; samplesPerSecond = 10429.9
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.19775096 * 276; EvalClassificationError = 0.58333333 * 276; time = 0.0111s; samplesPerSecond = 24950.3
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 2.14623378 * 260; EvalClassificationError = 0.58076923 * 260; time = 0.0109s; samplesPerSecond = 23903.6
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.17 seconds since last report (0.00 seconds on comm.); 4253 samples processed by 2 workers (2073 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 25.18k samplesPerSecond , throughputPerWorker = 12.59k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.98115282 * 265; EvalErrorPrediction = 0.54716981 * 265; time = 0.0346s; samplesPerSecond = 7651.2
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.13033706 * 298; EvalErrorPrediction = 0.57718121 * 298; time = 0.0225s; samplesPerSecond = 13269.8
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.96671704 * 274; EvalErrorPrediction = 0.50364964 * 274; time = 0.0210s; samplesPerSecond = 13072.5
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.98515695 * 265; EvalErrorPrediction = 0.53962264 * 265; time = 0.0247s; samplesPerSecond = 10747.5
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.85824670 * 281; EvalErrorPrediction = 0.54448399 * 281; time = 0.0211s; samplesPerSecond = 13300.5
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.95611759 * 252; EvalErrorPrediction = 0.52777778 * 252; time = 0.0162s; samplesPerSecond = 15546.0
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 2.04102834 * 274; EvalErrorPrediction = 0.51094891 * 274; time = 0.0207s; samplesPerSecond = 13237.4
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.84439304 * 258; EvalErrorPrediction = 0.49224806 * 258; time = 0.0108s; samplesPerSecond = 23807.3
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.98115282 * 265; EvalClassificationError = 0.54716981 * 265; time = 0.0346s; samplesPerSecond = 7651.2
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.13033706 * 298; EvalClassificationError = 0.57718121 * 298; time = 0.0225s; samplesPerSecond = 13269.8
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.96671704 * 274; EvalClassificationError = 0.50364964 * 274; time = 0.0210s; samplesPerSecond = 13072.5
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.98515695 * 265; EvalClassificationError = 0.53962264 * 265; time = 0.0247s; samplesPerSecond = 10747.5
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.85824670 * 281; EvalClassificationError = 0.54448399 * 281; time = 0.0211s; samplesPerSecond = 13300.5
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.95611759 * 252; EvalClassificationError = 0.52777778 * 252; time = 0.0162s; samplesPerSecond = 15546.0
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 2.04102834 * 274; EvalClassificationError = 0.51094891 * 274; time = 0.0207s; samplesPerSecond = 13237.4
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.84439304 * 258; EvalClassificationError = 0.49224806 * 258; time = 0.0108s; samplesPerSecond = 23807.3
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.17 seconds since last report (0.00 seconds on comm.); 4246 samples processed by 2 workers (2102 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 25.14k samplesPerSecond , throughputPerWorker = 12.57k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 1.97159866 * 271; EvalErrorPrediction = 0.55719557 * 271; time = 0.0357s; samplesPerSecond = 7592.3
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.97632511 * 278; EvalErrorPrediction = 0.52877698 * 278; time = 0.0248s; samplesPerSecond = 11196.1
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.95095600 * 286; EvalErrorPrediction = 0.56293706 * 286; time = 0.0238s; samplesPerSecond = 12010.2
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.94334189 * 170; EvalErrorPrediction = 0.52941176 * 170; time = 0.0156s; samplesPerSecond = 10894.6
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 1.97159866 * 271; EvalClassificationError = 0.55719557 * 271; time = 0.0357s; samplesPerSecond = 7592.3
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.97632511 * 278; EvalClassificationError = 0.52877698 * 278; time = 0.0248s; samplesPerSecond = 11196.1
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.95095600 * 286; EvalClassificationError = 0.56293706 * 286; time = 0.0238s; samplesPerSecond = 12010.2
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.94334189 * 170; EvalClassificationError = 0.52941176 * 170; time = 0.0156s; samplesPerSecond = 10894.6
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.27 seconds since last report (0.16 seconds on comm.); 7692 samples processed by 2 workers (904 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 28.58k samplesPerSecond , throughputPerWorker = 14.29k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.05172118 * 20480; EvalErrorPrediction = 0.55805664 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.785031s
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.05172118 * 20480; EvalClassificationError = 0.55805664 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.785031s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.94144328 * 1130; EvalErrorPrediction = 0.53097345 * 1130; time = 0.0754s; samplesPerSecond = 14987.7
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.94144328 * 1130; EvalClassificationError = 0.53097345 * 1130; time = 0.0754s; samplesPerSecond = 14987.7
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.13 seconds since last report (0.00 seconds on comm.); 4885 samples processed by 2 workers (2293 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 36.49k samplesPerSecond , throughputPerWorker = 18.25k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.87694957 * 1163; EvalErrorPrediction = 0.52794497 * 1163; time = 0.0547s; samplesPerSecond = 21257.9
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.98228580 * 1085; EvalErrorPrediction = 0.52718894 * 1085; time = 0.0548s; samplesPerSecond = 19810.8
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.87694957 * 1163; EvalClassificationError = 0.52794497 * 1163; time = 0.0547s; samplesPerSecond = 21257.9
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.98228580 * 1085; EvalClassificationError = 0.52718894 * 1085; time = 0.0548s; samplesPerSecond = 19810.8
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.12 seconds since last report (0.01 seconds on comm.); 4826 samples processed by 2 workers (2249 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 39.11k samplesPerSecond , throughputPerWorker = 19.55k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.92865528 * 1164; EvalErrorPrediction = 0.54725086 * 1164; time = 0.0685s; samplesPerSecond = 16995.7
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 2.00184241 * 1167; EvalErrorPrediction = 0.57155099 * 1167; time = 0.0574s; samplesPerSecond = 20343.4
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.92865528 * 1164; EvalClassificationError = 0.54725086 * 1164; time = 0.0685s; samplesPerSecond = 16995.7
|
||||
MPI Rank 1: 08/16/2016 10:09:00: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 2.00184241 * 1167; EvalClassificationError = 0.57155099 * 1167; time = 0.0574s; samplesPerSecond = 20343.4
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.12 seconds since last report (0.00 seconds on comm.); 4903 samples processed by 2 workers (2326 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 40.04k samplesPerSecond , throughputPerWorker = 20.02k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.92549525 * 1159; EvalErrorPrediction = 0.54270923 * 1159; time = 0.0649s; samplesPerSecond = 17851.9
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.98011842 * 823; EvalErrorPrediction = 0.54921021 * 823; time = 0.0592s; samplesPerSecond = 13905.1
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.92549525 * 1159; EvalClassificationError = 0.54270923 * 1159; time = 0.0649s; samplesPerSecond = 17851.9
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.98011842 * 823; EvalClassificationError = 0.54921021 * 823; time = 0.0592s; samplesPerSecond = 13905.1
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.01-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.16 seconds since last report (0.07 seconds on comm.); 5866 samples processed by 2 workers (823 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 37.73k samplesPerSecond , throughputPerWorker = 18.86k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.95703393 * 20480; EvalErrorPrediction = 0.54541016 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.535498s
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.95703393 * 20480; EvalClassificationError = 0.54541016 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.535498s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.89485201 * 1149; EvalErrorPrediction = 0.53176675 * 1149; time = 0.0577s; samplesPerSecond = 19921.6
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.89485201 * 1149; EvalClassificationError = 0.53176675 * 1149; time = 0.0577s; samplesPerSecond = 19921.6
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.13 seconds since last report (0.00 seconds on comm.); 4901 samples processed by 2 workers (2351 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 38.06k samplesPerSecond , throughputPerWorker = 19.03k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94072250 * 1202; EvalErrorPrediction = 0.53826955 * 1202; time = 0.0679s; samplesPerSecond = 17708.2
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.90064937 * 1130; EvalErrorPrediction = 0.52300885 * 1130; time = 0.0597s; samplesPerSecond = 18924.5
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94072250 * 1202; EvalClassificationError = 0.53826955 * 1202; time = 0.0679s; samplesPerSecond = 17708.2
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.90064937 * 1130; EvalClassificationError = 0.52300885 * 1130; time = 0.0597s; samplesPerSecond = 18924.5
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.12 seconds since last report (0.00 seconds on comm.); 4836 samples processed by 2 workers (2317 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 40.33k samplesPerSecond , throughputPerWorker = 20.17k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.85577719 * 1187; EvalErrorPrediction = 0.50884583 * 1187; time = 0.0600s; samplesPerSecond = 19773.1
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.94097997 * 1202; EvalErrorPrediction = 0.54658902 * 1202; time = 0.0571s; samplesPerSecond = 21043.4
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.85577719 * 1187; EvalClassificationError = 0.50884583 * 1187; time = 0.0600s; samplesPerSecond = 19773.1
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.94097997 * 1202; EvalClassificationError = 0.54658902 * 1202; time = 0.0571s; samplesPerSecond = 21043.4
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.12 seconds since last report (0.00 seconds on comm.); 4952 samples processed by 2 workers (2401 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 40.08k samplesPerSecond , throughputPerWorker = 20.04k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.91747174 * 1199; EvalErrorPrediction = 0.54211843 * 1199; time = 0.0663s; samplesPerSecond = 18092.9
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.91766783 * 817; EvalErrorPrediction = 0.53243574 * 817; time = 0.0456s; samplesPerSecond = 17920.6
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.91747174 * 1199; EvalClassificationError = 0.54211843 * 1199; time = 0.0663s; samplesPerSecond = 18092.9
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.91766783 * 817; EvalClassificationError = 0.53243574 * 817; time = 0.0456s; samplesPerSecond = 17920.6
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.14 seconds since last report (0.07 seconds on comm.); 5791 samples processed by 2 workers (817 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 40.15k samplesPerSecond , throughputPerWorker = 20.08k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.90306770 * 20480; EvalErrorPrediction = 0.52641602 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.516694s
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.90306770 * 20480; EvalClassificationError = 0.52641602 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.516694s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Starting Epoch 5: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 4: frames [81920..102400] (first utterance at frame 81920), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.86643684 * 1209; EvalErrorPrediction = 0.50372208 * 1209; time = 0.0642s; samplesPerSecond = 18842.9
|
||||
MPI Rank 1: 08/16/2016 10:09:01: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.86643684 * 1209; EvalClassificationError = 0.50372208 * 1209; time = 0.0642s; samplesPerSecond = 18842.9
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.14 seconds since last report (0.00 seconds on comm.); 4919 samples processed by 2 workers (2426 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 35.48k samplesPerSecond , throughputPerWorker = 17.74k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94031579 * 1217; EvalErrorPrediction = 0.53327855 * 1217; time = 0.0713s; samplesPerSecond = 17072.8
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.89794045 * 1206; EvalErrorPrediction = 0.51824212 * 1206; time = 0.0617s; samplesPerSecond = 19537.6
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94031579 * 1217; EvalClassificationError = 0.53327855 * 1217; time = 0.0713s; samplesPerSecond = 17072.8
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.89794045 * 1206; EvalClassificationError = 0.51824212 * 1206; time = 0.0617s; samplesPerSecond = 19537.6
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.13 seconds since last report (0.00 seconds on comm.); 4899 samples processed by 2 workers (2419 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 37.61k samplesPerSecond , throughputPerWorker = 18.80k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.96415395 * 1213; EvalErrorPrediction = 0.53833471 * 1213; time = 0.0684s; samplesPerSecond = 17741.7
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.84516499 * 1147; EvalErrorPrediction = 0.51264167 * 1147; time = 0.0547s; samplesPerSecond = 20981.6
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.96415395 * 1213; EvalClassificationError = 0.53833471 * 1213; time = 0.0684s; samplesPerSecond = 17741.7
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.84516499 * 1147; EvalClassificationError = 0.51264167 * 1147; time = 0.0547s; samplesPerSecond = 20981.6
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.13 seconds since last report (0.00 seconds on comm.); 4829 samples processed by 2 workers (2359 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 37.39k samplesPerSecond , throughputPerWorker = 18.70k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 2.02566421 * 1212; EvalErrorPrediction = 0.55280528 * 1212; time = 0.0743s; samplesPerSecond = 16309.6
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.89201183 * 809; EvalErrorPrediction = 0.50679852 * 809; time = 0.0369s; samplesPerSecond = 21937.8
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 2.02566421 * 1212; EvalClassificationError = 0.55280528 * 1212; time = 0.0743s; samplesPerSecond = 16309.6
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.89201183 * 809; EvalClassificationError = 0.50679852 * 809; time = 0.0369s; samplesPerSecond = 21937.8
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.03 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.15 seconds since last report (0.07 seconds on comm.); 5833 samples processed by 2 workers (809 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 37.93k samplesPerSecond , throughputPerWorker = 18.96k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.88963745 * 20480; EvalErrorPrediction = 0.51865234 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-05; epochTime=0.552104s
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.88963745 * 20480; EvalClassificationError = 0.51865234 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-05; epochTime=0.552104s
|
||||
MPI Rank 1: 08/16/2016 10:09:02: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:02: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -104,7 +104,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -130,7 +130,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -212,7 +212,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -238,7 +238,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -326,7 +326,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -352,7 +352,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -451,7 +451,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -480,7 +480,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -503,7 +503,7 @@ MPI Rank 0: 08/16/2016 03:19:40: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:19:40: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:40: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:19:40: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:19:40: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -556,113 +556,113 @@ MPI Rank 0: 08/16/2016 03:19:44: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.66660459 * 192; EvalErrorPrediction = 0.91666667 * 192; time = 0.0508s; samplesPerSecond = 3781.6
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.37978218 * 192; EvalErrorPrediction = 0.88541667 * 192; time = 0.0294s; samplesPerSecond = 6522.2
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.69675309 * 192; EvalErrorPrediction = 0.93229167 * 192; time = 0.0276s; samplesPerSecond = 6957.3
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.23088190 * 192; EvalErrorPrediction = 0.91666667 * 192; time = 0.0276s; samplesPerSecond = 6952.5
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.41262185 * 192; EvalErrorPrediction = 0.90625000 * 192; time = 0.0271s; samplesPerSecond = 7081.2
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.63%]: CrossEntropyWithSoftmax = 4.35726114 * 192; EvalErrorPrediction = 0.94270833 * 192; time = 0.0266s; samplesPerSecond = 7222.1
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.22315242 * 192; EvalErrorPrediction = 0.96354167 * 192; time = 0.0271s; samplesPerSecond = 7079.4
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06862554 * 192; EvalErrorPrediction = 0.89062500 * 192; time = 0.0276s; samplesPerSecond = 6967.6
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97387863 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0285s; samplesPerSecond = 6743.0
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83243560 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0290s; samplesPerSecond = 6619.8
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79996973 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0279s; samplesPerSecond = 6884.7
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.74380828 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0280s; samplesPerSecond = 6863.3
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73812888 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0265s; samplesPerSecond = 7236.5
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.13%]: CrossEntropyWithSoftmax = 3.69163791 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0278s; samplesPerSecond = 6902.5
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.88173262 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0276s; samplesPerSecond = 6963.1
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.77142322 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0288s; samplesPerSecond = 6664.4
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81063177 * 192; EvalErrorPrediction = 0.88020833 * 192; time = 0.0277s; samplesPerSecond = 6932.7
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.91374076 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0281s; samplesPerSecond = 6841.3
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.55613122 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0270s; samplesPerSecond = 7115.3
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.64148239 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0271s; samplesPerSecond = 7089.1
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.35544611 * 192; EvalErrorPrediction = 0.75520833 * 192; time = 0.0301s; samplesPerSecond = 6379.4
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.63%]: CrossEntropyWithSoftmax = 3.40034303 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0273s; samplesPerSecond = 7044.6
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.34712754 * 192; EvalErrorPrediction = 0.76041667 * 192; time = 0.0286s; samplesPerSecond = 6713.8
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.66023382 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0296s; samplesPerSecond = 6481.7
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.62838263 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0283s; samplesPerSecond = 6774.9
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.36560063 * 192; EvalErrorPrediction = 0.83854167 * 192; time = 0.0286s; samplesPerSecond = 6711.4
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.50734757 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0276s; samplesPerSecond = 6959.3
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47359624 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0274s; samplesPerSecond = 7015.7
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.55650851 * 192; EvalErrorPrediction = 0.81250000 * 192; time = 0.0268s; samplesPerSecond = 7165.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.13%]: CrossEntropyWithSoftmax = 3.44003656 * 192; EvalErrorPrediction = 0.80208333 * 192; time = 0.0276s; samplesPerSecond = 6959.0
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.35749237 * 192; EvalErrorPrediction = 0.77083333 * 192; time = 0.0280s; samplesPerSecond = 6845.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55640565 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0274s; samplesPerSecond = 7013.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.32958052 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0279s; samplesPerSecond = 6882.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.50833357 * 192; EvalErrorPrediction = 0.84375000 * 192; time = 0.0282s; samplesPerSecond = 6813.6
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.54460465 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0283s; samplesPerSecond = 6788.5
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.47456552 * 192; EvalErrorPrediction = 0.81250000 * 192; time = 0.0275s; samplesPerSecond = 6992.8
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.31118826 * 192; EvalErrorPrediction = 0.79687500 * 192; time = 0.0271s; samplesPerSecond = 7074.7
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.63%]: CrossEntropyWithSoftmax = 3.23043001 * 192; EvalErrorPrediction = 0.78645833 * 192; time = 0.0273s; samplesPerSecond = 7027.3
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.18504533 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0272s; samplesPerSecond = 7052.6
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23396798 * 192; EvalErrorPrediction = 0.74479167 * 192; time = 0.0261s; samplesPerSecond = 7368.5
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.51781736 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0290s; samplesPerSecond = 6610.0
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 2.94438285 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0269s; samplesPerSecond = 7133.6
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.05991271 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0262s; samplesPerSecond = 7338.0
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.02642719 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0262s; samplesPerSecond = 7316.5
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.13208851 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0266s; samplesPerSecond = 7206.4
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.13%]: CrossEntropyWithSoftmax = 2.72670959 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0275s; samplesPerSecond = 6981.8
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.12385008 * 192; EvalErrorPrediction = 0.80208333 * 192; time = 0.0264s; samplesPerSecond = 7263.9
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 2.98203627 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0270s; samplesPerSecond = 7114.5
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 2.97060074 * 192; EvalErrorPrediction = 0.75520833 * 192; time = 0.0286s; samplesPerSecond = 6724.3
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.74938235 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0292s; samplesPerSecond = 6574.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.02481319 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0274s; samplesPerSecond = 6994.5
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.11357602 * 192; EvalErrorPrediction = 0.76041667 * 192; time = 0.0280s; samplesPerSecond = 6858.1
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.07046118 * 192; EvalErrorPrediction = 0.74479167 * 192; time = 0.0272s; samplesPerSecond = 7047.4
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.63%]: CrossEntropyWithSoftmax = 2.80324721 * 192; EvalErrorPrediction = 0.72395833 * 192; time = 0.0268s; samplesPerSecond = 7170.1
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.74425310 * 192; EvalErrorPrediction = 0.68750000 * 192; time = 0.0275s; samplesPerSecond = 6983.8
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 2.95508757 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0272s; samplesPerSecond = 7047.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69285817 * 192; EvalErrorPrediction = 0.67708333 * 192; time = 0.0268s; samplesPerSecond = 7154.6
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.61222587 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0280s; samplesPerSecond = 6863.0
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.65528671 * 192; EvalErrorPrediction = 0.65625000 * 192; time = 0.0298s; samplesPerSecond = 6446.6
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.68470586 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0282s; samplesPerSecond = 6807.8
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.78009167 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0296s; samplesPerSecond = 6492.0
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.75177433 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0291s; samplesPerSecond = 6607.2
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.71744614 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0272s; samplesPerSecond = 7047.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.78208591 * 192; EvalErrorPrediction = 0.67708333 * 192; time = 0.0290s; samplesPerSecond = 6612.5
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.50297451 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0268s; samplesPerSecond = 7159.1
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97786137 * 192; EvalErrorPrediction = 0.75520833 * 192; time = 0.0277s; samplesPerSecond = 6920.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.60160999 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0264s; samplesPerSecond = 7263.1
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.49883465 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0277s; samplesPerSecond = 6939.7
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.58102011 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0271s; samplesPerSecond = 7088.5
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.63%]: CrossEntropyWithSoftmax = 2.55279515 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0270s; samplesPerSecond = 7102.4
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.55386415 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0276s; samplesPerSecond = 6959.3
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.44782705 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0277s; samplesPerSecond = 6927.2
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.71017278 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0266s; samplesPerSecond = 7219.4
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.48043969 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0269s; samplesPerSecond = 7142.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.41207736 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0273s; samplesPerSecond = 7024.0
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.71955135 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0281s; samplesPerSecond = 6839.8
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.52658962 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0279s; samplesPerSecond = 6877.3
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.13%]: CrossEntropyWithSoftmax = 2.95080890 * 192; EvalErrorPrediction = 0.77083333 * 192; time = 0.0276s; samplesPerSecond = 6968.1
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.32160400 * 192; EvalErrorPrediction = 0.57812500 * 192; time = 0.0310s; samplesPerSecond = 6195.3
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.21126458 * 192; EvalErrorPrediction = 0.61979167 * 192; time = 0.0276s; samplesPerSecond = 6967.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.28277798 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0281s; samplesPerSecond = 6827.6
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.66166449 * 192; EvalErrorPrediction = 0.68750000 * 192; time = 0.0292s; samplesPerSecond = 6580.8
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.41115800 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0272s; samplesPerSecond = 7063.0
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.47460061 * 192; EvalErrorPrediction = 0.65625000 * 192; time = 0.0263s; samplesPerSecond = 7300.1
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.35103706 * 192; EvalErrorPrediction = 0.61979167 * 192; time = 0.0277s; samplesPerSecond = 6939.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.63%]: CrossEntropyWithSoftmax = 2.33610739 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0278s; samplesPerSecond = 6909.7
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.29476428 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0270s; samplesPerSecond = 7111.4
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.25023789 * 192; EvalErrorPrediction = 0.59375000 * 192; time = 0.0273s; samplesPerSecond = 7025.8
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.03832213 * 192; EvalErrorPrediction = 0.51041667 * 192; time = 0.0269s; samplesPerSecond = 7147.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.37205200 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0271s; samplesPerSecond = 7073.1
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.50009905 * 192; EvalErrorPrediction = 0.65625000 * 192; time = 0.0274s; samplesPerSecond = 6998.4
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.32391497 * 192; EvalErrorPrediction = 0.61979167 * 192; time = 0.0287s; samplesPerSecond = 6678.5
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.56896275 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0266s; samplesPerSecond = 7216.1
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.13%]: CrossEntropyWithSoftmax = 2.44517270 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0315s; samplesPerSecond = 6087.9
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.26122471 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0275s; samplesPerSecond = 6985.6
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.17377959 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0269s; samplesPerSecond = 7143.7
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.22791977 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0280s; samplesPerSecond = 6848.1
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.07889758 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0267s; samplesPerSecond = 7191.0
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.27676510 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0270s; samplesPerSecond = 7104.5
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26006065 * 192; EvalErrorPrediction = 0.60937500 * 192; time = 0.0285s; samplesPerSecond = 6737.1
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.11239689 * 192; EvalErrorPrediction = 0.56250000 * 192; time = 0.0269s; samplesPerSecond = 7137.3
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.63%]: CrossEntropyWithSoftmax = 2.24244492 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0243s; samplesPerSecond = 7907.7
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.14233722 * 192; EvalErrorPrediction = 0.54687500 * 192; time = 0.0242s; samplesPerSecond = 7933.9
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.40359040 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0247s; samplesPerSecond = 7780.2
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.14246064 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0249s; samplesPerSecond = 7709.3
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.18541632 * 192; EvalErrorPrediction = 0.57812500 * 192; time = 0.0219s; samplesPerSecond = 8773.1
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.03288s
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.66660459 * 192; EvalClassificationError = 0.91666667 * 192; time = 0.0508s; samplesPerSecond = 3781.6
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.37978218 * 192; EvalClassificationError = 0.88541667 * 192; time = 0.0294s; samplesPerSecond = 6522.2
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.69675309 * 192; EvalClassificationError = 0.93229167 * 192; time = 0.0276s; samplesPerSecond = 6957.3
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.23088190 * 192; EvalClassificationError = 0.91666667 * 192; time = 0.0276s; samplesPerSecond = 6952.5
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.41262185 * 192; EvalClassificationError = 0.90625000 * 192; time = 0.0271s; samplesPerSecond = 7081.2
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.63%]: CrossEntropyWithSoftmax = 4.35726114 * 192; EvalClassificationError = 0.94270833 * 192; time = 0.0266s; samplesPerSecond = 7222.1
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.22315242 * 192; EvalClassificationError = 0.96354167 * 192; time = 0.0271s; samplesPerSecond = 7079.4
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06862554 * 192; EvalClassificationError = 0.89062500 * 192; time = 0.0276s; samplesPerSecond = 6967.6
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97387863 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0285s; samplesPerSecond = 6743.0
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83243560 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0290s; samplesPerSecond = 6619.8
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79996973 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0279s; samplesPerSecond = 6884.7
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.74380828 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0280s; samplesPerSecond = 6863.3
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73812888 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0265s; samplesPerSecond = 7236.5
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.13%]: CrossEntropyWithSoftmax = 3.69163791 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0278s; samplesPerSecond = 6902.5
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.88173262 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0276s; samplesPerSecond = 6963.1
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.77142322 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0288s; samplesPerSecond = 6664.4
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81063177 * 192; EvalClassificationError = 0.88020833 * 192; time = 0.0277s; samplesPerSecond = 6932.7
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.91374076 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0281s; samplesPerSecond = 6841.3
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.55613122 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0270s; samplesPerSecond = 7115.3
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.64148239 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0271s; samplesPerSecond = 7089.1
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.35544611 * 192; EvalClassificationError = 0.75520833 * 192; time = 0.0301s; samplesPerSecond = 6379.4
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.63%]: CrossEntropyWithSoftmax = 3.40034303 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0273s; samplesPerSecond = 7044.6
|
||||
MPI Rank 0: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.34712754 * 192; EvalClassificationError = 0.76041667 * 192; time = 0.0286s; samplesPerSecond = 6713.8
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.66023382 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0296s; samplesPerSecond = 6481.7
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.62838263 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0283s; samplesPerSecond = 6774.9
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.36560063 * 192; EvalClassificationError = 0.83854167 * 192; time = 0.0286s; samplesPerSecond = 6711.4
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.50734757 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0276s; samplesPerSecond = 6959.3
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47359624 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0274s; samplesPerSecond = 7015.7
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.55650851 * 192; EvalClassificationError = 0.81250000 * 192; time = 0.0268s; samplesPerSecond = 7165.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.13%]: CrossEntropyWithSoftmax = 3.44003656 * 192; EvalClassificationError = 0.80208333 * 192; time = 0.0276s; samplesPerSecond = 6959.0
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.35749237 * 192; EvalClassificationError = 0.77083333 * 192; time = 0.0280s; samplesPerSecond = 6845.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55640565 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0274s; samplesPerSecond = 7013.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.32958052 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0279s; samplesPerSecond = 6882.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.50833357 * 192; EvalClassificationError = 0.84375000 * 192; time = 0.0282s; samplesPerSecond = 6813.6
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.54460465 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0283s; samplesPerSecond = 6788.5
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.47456552 * 192; EvalClassificationError = 0.81250000 * 192; time = 0.0275s; samplesPerSecond = 6992.8
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.31118826 * 192; EvalClassificationError = 0.79687500 * 192; time = 0.0271s; samplesPerSecond = 7074.7
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.63%]: CrossEntropyWithSoftmax = 3.23043001 * 192; EvalClassificationError = 0.78645833 * 192; time = 0.0273s; samplesPerSecond = 7027.3
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.18504533 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0272s; samplesPerSecond = 7052.6
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23396798 * 192; EvalClassificationError = 0.74479167 * 192; time = 0.0261s; samplesPerSecond = 7368.5
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.51781736 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0290s; samplesPerSecond = 6610.0
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 2.94438285 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0269s; samplesPerSecond = 7133.6
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.05991271 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0262s; samplesPerSecond = 7338.0
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.02642719 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0262s; samplesPerSecond = 7316.5
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.13208851 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0266s; samplesPerSecond = 7206.4
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.13%]: CrossEntropyWithSoftmax = 2.72670959 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0275s; samplesPerSecond = 6981.8
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.12385008 * 192; EvalClassificationError = 0.80208333 * 192; time = 0.0264s; samplesPerSecond = 7263.9
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 2.98203627 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0270s; samplesPerSecond = 7114.5
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 2.97060074 * 192; EvalClassificationError = 0.75520833 * 192; time = 0.0286s; samplesPerSecond = 6724.3
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.74938235 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0292s; samplesPerSecond = 6574.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.02481319 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0274s; samplesPerSecond = 6994.5
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.11357602 * 192; EvalClassificationError = 0.76041667 * 192; time = 0.0280s; samplesPerSecond = 6858.1
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.07046118 * 192; EvalClassificationError = 0.74479167 * 192; time = 0.0272s; samplesPerSecond = 7047.4
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.63%]: CrossEntropyWithSoftmax = 2.80324721 * 192; EvalClassificationError = 0.72395833 * 192; time = 0.0268s; samplesPerSecond = 7170.1
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.74425310 * 192; EvalClassificationError = 0.68750000 * 192; time = 0.0275s; samplesPerSecond = 6983.8
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 2.95508757 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0272s; samplesPerSecond = 7047.2
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69285817 * 192; EvalClassificationError = 0.67708333 * 192; time = 0.0268s; samplesPerSecond = 7154.6
|
||||
MPI Rank 0: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.61222587 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0280s; samplesPerSecond = 6863.0
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.65528671 * 192; EvalClassificationError = 0.65625000 * 192; time = 0.0298s; samplesPerSecond = 6446.6
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.68470586 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0282s; samplesPerSecond = 6807.8
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.78009167 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0296s; samplesPerSecond = 6492.0
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.75177433 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0291s; samplesPerSecond = 6607.2
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.71744614 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0272s; samplesPerSecond = 7047.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.78208591 * 192; EvalClassificationError = 0.67708333 * 192; time = 0.0290s; samplesPerSecond = 6612.5
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.50297451 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0268s; samplesPerSecond = 7159.1
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97786137 * 192; EvalClassificationError = 0.75520833 * 192; time = 0.0277s; samplesPerSecond = 6920.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.60160999 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0264s; samplesPerSecond = 7263.1
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.49883465 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0277s; samplesPerSecond = 6939.7
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.58102011 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0271s; samplesPerSecond = 7088.5
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.63%]: CrossEntropyWithSoftmax = 2.55279515 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0270s; samplesPerSecond = 7102.4
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.55386415 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0276s; samplesPerSecond = 6959.3
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.44782705 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0277s; samplesPerSecond = 6927.2
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.71017278 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0266s; samplesPerSecond = 7219.4
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.48043969 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0269s; samplesPerSecond = 7142.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.41207736 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0273s; samplesPerSecond = 7024.0
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.71955135 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0281s; samplesPerSecond = 6839.8
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.52658962 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0279s; samplesPerSecond = 6877.3
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.13%]: CrossEntropyWithSoftmax = 2.95080890 * 192; EvalClassificationError = 0.77083333 * 192; time = 0.0276s; samplesPerSecond = 6968.1
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.32160400 * 192; EvalClassificationError = 0.57812500 * 192; time = 0.0310s; samplesPerSecond = 6195.3
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.21126458 * 192; EvalClassificationError = 0.61979167 * 192; time = 0.0276s; samplesPerSecond = 6967.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.28277798 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0281s; samplesPerSecond = 6827.6
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.66166449 * 192; EvalClassificationError = 0.68750000 * 192; time = 0.0292s; samplesPerSecond = 6580.8
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.41115800 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0272s; samplesPerSecond = 7063.0
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.47460061 * 192; EvalClassificationError = 0.65625000 * 192; time = 0.0263s; samplesPerSecond = 7300.1
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.35103706 * 192; EvalClassificationError = 0.61979167 * 192; time = 0.0277s; samplesPerSecond = 6939.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.63%]: CrossEntropyWithSoftmax = 2.33610739 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0278s; samplesPerSecond = 6909.7
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.29476428 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0270s; samplesPerSecond = 7111.4
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.25023789 * 192; EvalClassificationError = 0.59375000 * 192; time = 0.0273s; samplesPerSecond = 7025.8
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.03832213 * 192; EvalClassificationError = 0.51041667 * 192; time = 0.0269s; samplesPerSecond = 7147.9
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.37205200 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0271s; samplesPerSecond = 7073.1
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.50009905 * 192; EvalClassificationError = 0.65625000 * 192; time = 0.0274s; samplesPerSecond = 6998.4
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.32391497 * 192; EvalClassificationError = 0.61979167 * 192; time = 0.0287s; samplesPerSecond = 6678.5
|
||||
MPI Rank 0: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.56896275 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0266s; samplesPerSecond = 7216.1
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.13%]: CrossEntropyWithSoftmax = 2.44517270 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0315s; samplesPerSecond = 6087.9
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.26122471 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0275s; samplesPerSecond = 6985.6
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.17377959 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0269s; samplesPerSecond = 7143.7
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.22791977 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0280s; samplesPerSecond = 6848.1
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.07889758 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0267s; samplesPerSecond = 7191.0
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.27676510 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0270s; samplesPerSecond = 7104.5
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26006065 * 192; EvalClassificationError = 0.60937500 * 192; time = 0.0285s; samplesPerSecond = 6737.1
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.11239689 * 192; EvalClassificationError = 0.56250000 * 192; time = 0.0269s; samplesPerSecond = 7137.3
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.63%]: CrossEntropyWithSoftmax = 2.24244492 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0243s; samplesPerSecond = 7907.7
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.14233722 * 192; EvalClassificationError = 0.54687500 * 192; time = 0.0242s; samplesPerSecond = 7933.9
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.40359040 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0247s; samplesPerSecond = 7780.2
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.14246064 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0249s; samplesPerSecond = 7709.3
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.18541632 * 192; EvalClassificationError = 0.57812500 * 192; time = 0.0219s; samplesPerSecond = 8773.1
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.03288s
|
||||
MPI Rank 0: 08/16/2016 03:19:47: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBM@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -670,46 +670,46 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.16795139 * 508; EvalErrorPrediction = 0.58661417 * 508; time = 0.0803s; samplesPerSecond = 6330.1
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.15396933 * 492; EvalErrorPrediction = 0.57520325 * 492; time = 0.0617s; samplesPerSecond = 7975.9
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.13127278 * 488; EvalErrorPrediction = 0.59426230 * 488; time = 0.0545s; samplesPerSecond = 8958.2
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.19725263 * 527; EvalErrorPrediction = 0.57305503 * 527; time = 0.0664s; samplesPerSecond = 7942.4
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.16795139 * 508; EvalClassificationError = 0.58661417 * 508; time = 0.0803s; samplesPerSecond = 6330.1
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.15396933 * 492; EvalClassificationError = 0.57520325 * 492; time = 0.0617s; samplesPerSecond = 7975.9
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.13127278 * 488; EvalClassificationError = 0.59426230 * 488; time = 0.0545s; samplesPerSecond = 8958.2
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.19725263 * 527; EvalClassificationError = 0.57305503 * 527; time = 0.0664s; samplesPerSecond = 7942.4
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.01-seconds latency this time; accumulated time on sync point = 0.01 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.31 seconds since last report (0.00 seconds on comm.); 4289 samples processed by 2 workers (2163 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 13.80k samplesPerSecond , throughputPerWorker = 6.90k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 1.96790556 * 473; EvalErrorPrediction = 0.52219873 * 473; time = 0.0743s; samplesPerSecond = 6362.7
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.05563801 * 511; EvalErrorPrediction = 0.55772994 * 511; time = 0.0576s; samplesPerSecond = 8867.8
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.04775811 * 506; EvalErrorPrediction = 0.54545455 * 506; time = 0.0538s; samplesPerSecond = 9405.9
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.07825515 * 513; EvalErrorPrediction = 0.53021442 * 513; time = 0.0583s; samplesPerSecond = 8798.6
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 1.96790556 * 473; EvalClassificationError = 0.52219873 * 473; time = 0.0743s; samplesPerSecond = 6362.7
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.05563801 * 511; EvalClassificationError = 0.55772994 * 511; time = 0.0576s; samplesPerSecond = 8867.8
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.04775811 * 506; EvalClassificationError = 0.54545455 * 506; time = 0.0538s; samplesPerSecond = 9405.9
|
||||
MPI Rank 0: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.07825515 * 513; EvalClassificationError = 0.53021442 * 513; time = 0.0583s; samplesPerSecond = 8798.6
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.03 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.29 seconds since last report (0.00 seconds on comm.); 4253 samples processed by 2 workers (2180 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 14.86k samplesPerSecond , throughputPerWorker = 7.43k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.03188777 * 489; EvalErrorPrediction = 0.54805726 * 489; time = 0.0943s; samplesPerSecond = 5188.3
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.04744036 * 494; EvalErrorPrediction = 0.54251012 * 494; time = 0.0525s; samplesPerSecond = 9416.5
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 2.08216113 * 499; EvalErrorPrediction = 0.58517034 * 499; time = 0.0581s; samplesPerSecond = 8583.6
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.02506926 * 490; EvalErrorPrediction = 0.55102041 * 490; time = 0.0578s; samplesPerSecond = 8472.4
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.03188777 * 489; EvalClassificationError = 0.54805726 * 489; time = 0.0943s; samplesPerSecond = 5188.3
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.04744036 * 494; EvalClassificationError = 0.54251012 * 494; time = 0.0525s; samplesPerSecond = 9416.5
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 2.08216113 * 499; EvalClassificationError = 0.58517034 * 499; time = 0.0581s; samplesPerSecond = 8583.6
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.02506926 * 490; EvalClassificationError = 0.55102041 * 490; time = 0.0578s; samplesPerSecond = 8472.4
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.05 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.28 seconds since last report (0.00 seconds on comm.); 4246 samples processed by 2 workers (2144 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 15.11k samplesPerSecond , throughputPerWorker = 7.55k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 1.89488530 * 497; EvalErrorPrediction = 0.51106640 * 497; time = 0.0916s; samplesPerSecond = 5425.5
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.09344622 * 492; EvalErrorPrediction = 0.58333333 * 492; time = 0.0588s; samplesPerSecond = 8366.4
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 1.87997630 * 508; EvalErrorPrediction = 0.54133858 * 508; time = 0.0544s; samplesPerSecond = 9338.1
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.93826531 * 503; EvalErrorPrediction = 0.52485089 * 503; time = 0.0495s; samplesPerSecond = 10154.8
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.03991731 * 470; EvalErrorPrediction = 0.56382979 * 470; time = 0.0504s; samplesPerSecond = 9321.9
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.88433391 * 494; EvalErrorPrediction = 0.53846154 * 494; time = 0.0452s; samplesPerSecond = 10933.6
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.96044166 * 503; EvalErrorPrediction = 0.50894632 * 503; time = 0.0458s; samplesPerSecond = 10986.9
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.92271104 * 487; EvalErrorPrediction = 0.53798768 * 487; time = 0.0456s; samplesPerSecond = 10669.8
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.92298619 * 516; EvalErrorPrediction = 0.51744186 * 516; time = 0.0435s; samplesPerSecond = 11862.6
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 1.92695300 * 494; EvalErrorPrediction = 0.54858300 * 494; time = 0.0378s; samplesPerSecond = 13053.9
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.97859657 * 510; EvalErrorPrediction = 0.53725490 * 510; time = 0.0397s; samplesPerSecond = 12831.8
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 1.97569623 * 497; EvalErrorPrediction = 0.52917505 * 497; time = 0.0401s; samplesPerSecond = 12396.2
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.87129382 * 490; EvalErrorPrediction = 0.55714286 * 490; time = 0.0380s; samplesPerSecond = 12902.2
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.91961237 * 482; EvalErrorPrediction = 0.51452282 * 482; time = 0.0390s; samplesPerSecond = 12349.5
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.94802476 * 342; EvalErrorPrediction = 0.50877193 * 342; time = 0.0246s; samplesPerSecond = 13904.1
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 1.89488530 * 497; EvalClassificationError = 0.51106640 * 497; time = 0.0916s; samplesPerSecond = 5425.5
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.09344622 * 492; EvalClassificationError = 0.58333333 * 492; time = 0.0588s; samplesPerSecond = 8366.4
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 1.87997630 * 508; EvalClassificationError = 0.54133858 * 508; time = 0.0544s; samplesPerSecond = 9338.1
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.93826531 * 503; EvalClassificationError = 0.52485089 * 503; time = 0.0495s; samplesPerSecond = 10154.8
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.03991731 * 470; EvalClassificationError = 0.56382979 * 470; time = 0.0504s; samplesPerSecond = 9321.9
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.88433391 * 494; EvalClassificationError = 0.53846154 * 494; time = 0.0452s; samplesPerSecond = 10933.6
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.96044166 * 503; EvalClassificationError = 0.50894632 * 503; time = 0.0458s; samplesPerSecond = 10986.9
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.92271104 * 487; EvalClassificationError = 0.53798768 * 487; time = 0.0456s; samplesPerSecond = 10669.8
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.92298619 * 516; EvalClassificationError = 0.51744186 * 516; time = 0.0435s; samplesPerSecond = 11862.6
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 1.92695300 * 494; EvalClassificationError = 0.54858300 * 494; time = 0.0378s; samplesPerSecond = 13053.9
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.97859657 * 510; EvalClassificationError = 0.53725490 * 510; time = 0.0397s; samplesPerSecond = 12831.8
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 1.97569623 * 497; EvalClassificationError = 0.52917505 * 497; time = 0.0401s; samplesPerSecond = 12396.2
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.87129382 * 490; EvalClassificationError = 0.55714286 * 490; time = 0.0380s; samplesPerSecond = 12902.2
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.91961237 * 482; EvalClassificationError = 0.51452282 * 482; time = 0.0390s; samplesPerSecond = 12349.5
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.94802476 * 342; EvalClassificationError = 0.50877193 * 342; time = 0.0246s; samplesPerSecond = 13904.1
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.05 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.64 seconds since last report (0.00 seconds on comm.); 7692 samples processed by 2 workers (6788 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 12.04k samplesPerSecond , throughputPerWorker = 6.02k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.01405635 * 20480; EvalErrorPrediction = 0.55107422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.51892s
|
||||
MPI Rank 0: 08/16/2016 03:19:48: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.01405635 * 20480; EvalClassificationError = 0.55107422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.51892s
|
||||
MPI Rank 0: 08/16/2016 03:19:49: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBM@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:49: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -717,26 +717,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:49: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.92390973 * 1942; EvalErrorPrediction = 0.52523172 * 1942; time = 0.2104s; samplesPerSecond = 9230.2
|
||||
MPI Rank 0: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.92390973 * 1942; EvalClassificationError = 0.52523172 * 1942; time = 0.2104s; samplesPerSecond = 9230.2
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.31 seconds since last report (0.00 seconds on comm.); 4885 samples processed by 2 workers (2592 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 15.79k samplesPerSecond , throughputPerWorker = 7.89k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.91074269 * 1909; EvalErrorPrediction = 0.54216867 * 1909; time = 0.2074s; samplesPerSecond = 9206.6
|
||||
MPI Rank 0: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.91074269 * 1909; EvalClassificationError = 0.54216867 * 1909; time = 0.2074s; samplesPerSecond = 9206.6
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.28 seconds since last report (0.00 seconds on comm.); 4826 samples processed by 2 workers (2577 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 17.04k samplesPerSecond , throughputPerWorker = 8.52k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.95516876 * 1987; EvalErrorPrediction = 0.54755913 * 1987; time = 0.2380s; samplesPerSecond = 8347.8
|
||||
MPI Rank 0: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.95516876 * 1987; EvalClassificationError = 0.54755913 * 1987; time = 0.2380s; samplesPerSecond = 8347.8
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.29 seconds since last report (0.00 seconds on comm.); 4903 samples processed by 2 workers (2577 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 17.15k samplesPerSecond , throughputPerWorker = 8.58k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.90801486 * 1908; EvalErrorPrediction = 0.54402516 * 1908; time = 0.2143s; samplesPerSecond = 8902.9
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.94918099 * 1905; EvalErrorPrediction = 0.56220472 * 1905; time = 0.2000s; samplesPerSecond = 9523.7
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.93098232 * 1913; EvalErrorPrediction = 0.54417146 * 1913; time = 0.1513s; samplesPerSecond = 12642.5
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.94092937 * 1225; EvalErrorPrediction = 0.53959184 * 1225; time = 0.0931s; samplesPerSecond = 13152.2
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.90801486 * 1908; EvalClassificationError = 0.54402516 * 1908; time = 0.2143s; samplesPerSecond = 8902.9
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.94918099 * 1905; EvalClassificationError = 0.56220472 * 1905; time = 0.2000s; samplesPerSecond = 9523.7
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.93098232 * 1913; EvalClassificationError = 0.54417146 * 1913; time = 0.1513s; samplesPerSecond = 12642.5
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.94092937 * 1225; EvalClassificationError = 0.53959184 * 1225; time = 0.0931s; samplesPerSecond = 13152.2
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.46 seconds since last report (0.00 seconds on comm.); 5866 samples processed by 2 workers (5043 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 12.66k samplesPerSecond , throughputPerWorker = 6.33k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.92634052 * 20480; EvalErrorPrediction = 0.54199219 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.34331s
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.92634052 * 20480; EvalClassificationError = 0.54199219 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.34331s
|
||||
MPI Rank 0: 08/16/2016 03:19:50: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBM@release_cpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -744,26 +744,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.85973365 * 1923; EvalErrorPrediction = 0.51690068 * 1923; time = 0.1915s; samplesPerSecond = 10043.9
|
||||
MPI Rank 0: 08/16/2016 03:19:50: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.85973365 * 1923; EvalClassificationError = 0.51690068 * 1923; time = 0.1915s; samplesPerSecond = 10043.9
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.27 seconds since last report (0.00 seconds on comm.); 4901 samples processed by 2 workers (2550 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 17.86k samplesPerSecond , throughputPerWorker = 8.93k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.86650752 * 1870; EvalErrorPrediction = 0.52459893 * 1870; time = 0.1939s; samplesPerSecond = 9644.4
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.86650752 * 1870; EvalClassificationError = 0.52459893 * 1870; time = 0.1939s; samplesPerSecond = 9644.4
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.27 seconds since last report (0.00 seconds on comm.); 4836 samples processed by 2 workers (2519 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 17.97k samplesPerSecond , throughputPerWorker = 8.98k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.88231774 * 1942; EvalErrorPrediction = 0.53707518 * 1942; time = 0.2242s; samplesPerSecond = 8663.8
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.88231774 * 1942; EvalClassificationError = 0.53707518 * 1942; time = 0.2242s; samplesPerSecond = 8663.8
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.29 seconds since last report (0.00 seconds on comm.); 4952 samples processed by 2 workers (2551 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 17.07k samplesPerSecond , throughputPerWorker = 8.54k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.85291691 * 1885; EvalErrorPrediction = 0.51405836 * 1885; time = 0.2141s; samplesPerSecond = 8804.3
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.86882145 * 1870; EvalErrorPrediction = 0.50374332 * 1870; time = 0.1770s; samplesPerSecond = 10565.5
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.86819622 * 1873; EvalErrorPrediction = 0.52429258 * 1873; time = 0.1431s; samplesPerSecond = 13088.4
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.90127403 * 1231; EvalErrorPrediction = 0.51990252 * 1231; time = 0.0793s; samplesPerSecond = 15517.7
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.85291691 * 1885; EvalClassificationError = 0.51405836 * 1885; time = 0.2141s; samplesPerSecond = 8804.3
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.86882145 * 1870; EvalClassificationError = 0.50374332 * 1870; time = 0.1770s; samplesPerSecond = 10565.5
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.86819622 * 1873; EvalClassificationError = 0.52429258 * 1873; time = 0.1431s; samplesPerSecond = 13088.4
|
||||
MPI Rank 0: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.90127403 * 1231; EvalClassificationError = 0.51990252 * 1231; time = 0.0793s; samplesPerSecond = 15517.7
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.42 seconds since last report (0.00 seconds on comm.); 5791 samples processed by 2 workers (4974 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 13.84k samplesPerSecond , throughputPerWorker = 6.92k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:52: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.87410515 * 20480; EvalErrorPrediction = 0.52202148 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=1.25394s
|
||||
MPI Rank 0: 08/16/2016 03:19:52: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.87410515 * 20480; EvalClassificationError = 0.52202148 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=1.25394s
|
||||
MPI Rank 0: 08/16/2016 03:19:52: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBM@release_cpu/models/cntkSpeech.dnn.4'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:52: Starting Epoch 5: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -771,26 +771,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 4: frames [81920..102400] (first utterance at frame 81920), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:19:52: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.89785267 * 1863; EvalErrorPrediction = 0.52012882 * 1863; time = 0.1997s; samplesPerSecond = 9328.9
|
||||
MPI Rank 0: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.89785267 * 1863; EvalClassificationError = 0.52012882 * 1863; time = 0.1997s; samplesPerSecond = 9328.9
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.29 seconds since last report (0.00 seconds on comm.); 4919 samples processed by 2 workers (2493 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 16.99k samplesPerSecond , throughputPerWorker = 8.50k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.84186486 * 1855; EvalErrorPrediction = 0.51967655 * 1855; time = 0.2256s; samplesPerSecond = 8220.9
|
||||
MPI Rank 0: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.84186486 * 1855; EvalClassificationError = 0.51967655 * 1855; time = 0.2256s; samplesPerSecond = 8220.9
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.27 seconds since last report (0.00 seconds on comm.); 4899 samples processed by 2 workers (2480 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 17.86k samplesPerSecond , throughputPerWorker = 8.93k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.83137761 * 1866; EvalErrorPrediction = 0.51714898 * 1866; time = 0.1884s; samplesPerSecond = 9905.1
|
||||
MPI Rank 0: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.83137761 * 1866; EvalClassificationError = 0.51714898 * 1866; time = 0.1884s; samplesPerSecond = 9905.1
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.25 seconds since last report (0.00 seconds on comm.); 4829 samples processed by 2 workers (2470 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 19.18k samplesPerSecond , throughputPerWorker = 9.59k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.85137384 * 1859; EvalErrorPrediction = 0.52017214 * 1859; time = 0.1940s; samplesPerSecond = 9582.3
|
||||
MPI Rank 0: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.78938411 * 1925; EvalErrorPrediction = 0.48935065 * 1925; time = 0.1751s; samplesPerSecond = 10991.0
|
||||
MPI Rank 0: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.80988498 * 1860; EvalErrorPrediction = 0.49892473 * 1860; time = 0.1425s; samplesPerSecond = 13053.6
|
||||
MPI Rank 0: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.82839343 * 1239; EvalErrorPrediction = 0.50766747 * 1239; time = 0.0778s; samplesPerSecond = 15915.4
|
||||
MPI Rank 0: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.85137384 * 1859; EvalClassificationError = 0.52017214 * 1859; time = 0.1940s; samplesPerSecond = 9582.3
|
||||
MPI Rank 0: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.78938411 * 1925; EvalClassificationError = 0.48935065 * 1925; time = 0.1751s; samplesPerSecond = 10991.0
|
||||
MPI Rank 0: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.80988498 * 1860; EvalClassificationError = 0.49892473 * 1860; time = 0.1425s; samplesPerSecond = 13053.6
|
||||
MPI Rank 0: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.82839343 * 1239; EvalClassificationError = 0.50766747 * 1239; time = 0.0778s; samplesPerSecond = 15915.4
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.41 seconds since last report (0.00 seconds on comm.); 5833 samples processed by 2 workers (5024 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 14.07k samplesPerSecond , throughputPerWorker = 7.04k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:19:53: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.85694655 * 20480; EvalErrorPrediction = 0.51362305 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-005; epochTime=1.23173s
|
||||
MPI Rank 0: 08/16/2016 03:19:53: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.85694655 * 20480; EvalClassificationError = 0.51362305 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-005; epochTime=1.23173s
|
||||
MPI Rank 0: 08/16/2016 03:19:53: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBM@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:19:53: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -844,7 +844,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -870,7 +870,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -952,7 +952,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -978,7 +978,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1066,7 +1066,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1092,7 +1092,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1191,7 +1191,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1220,7 +1220,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1243,7 +1243,7 @@ MPI Rank 1: 08/16/2016 03:19:41: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:19:41: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:41: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:19:41: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:19:41: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1296,237 +1296,237 @@ MPI Rank 1: 08/16/2016 03:19:44: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.66660459 * 192; EvalErrorPrediction = 0.91666667 * 192; time = 0.0454s; samplesPerSecond = 4230.0
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.37978218 * 192; EvalErrorPrediction = 0.88541667 * 192; time = 0.0254s; samplesPerSecond = 7556.7
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.69675309 * 192; EvalErrorPrediction = 0.93229167 * 192; time = 0.0265s; samplesPerSecond = 7242.5
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.23088190 * 192; EvalErrorPrediction = 0.91666667 * 192; time = 0.0262s; samplesPerSecond = 7324.9
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.41262185 * 192; EvalErrorPrediction = 0.90625000 * 192; time = 0.0250s; samplesPerSecond = 7675.1
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.63%]: CrossEntropyWithSoftmax = 4.35726114 * 192; EvalErrorPrediction = 0.94270833 * 192; time = 0.0254s; samplesPerSecond = 7551.9
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.22315242 * 192; EvalErrorPrediction = 0.96354167 * 192; time = 0.0256s; samplesPerSecond = 7509.4
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06862554 * 192; EvalErrorPrediction = 0.89062500 * 192; time = 0.0258s; samplesPerSecond = 7441.6
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97387863 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0258s; samplesPerSecond = 7440.4
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83243560 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0275s; samplesPerSecond = 6988.9
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79996973 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0267s; samplesPerSecond = 7192.9
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.74380828 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0258s; samplesPerSecond = 7430.9
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73812888 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0248s; samplesPerSecond = 7741.3
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.13%]: CrossEntropyWithSoftmax = 3.69163791 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0254s; samplesPerSecond = 7548.1
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.88173262 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0261s; samplesPerSecond = 7370.4
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.77142322 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0252s; samplesPerSecond = 7610.0
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81063177 * 192; EvalErrorPrediction = 0.88020833 * 192; time = 0.0263s; samplesPerSecond = 7302.0
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.91374076 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0250s; samplesPerSecond = 7676.0
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.55613122 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0247s; samplesPerSecond = 7758.8
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.64148239 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0270s; samplesPerSecond = 7104.0
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.35544611 * 192; EvalErrorPrediction = 0.75520833 * 192; time = 0.0263s; samplesPerSecond = 7288.2
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.63%]: CrossEntropyWithSoftmax = 3.40034303 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0247s; samplesPerSecond = 7785.6
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.34712754 * 192; EvalErrorPrediction = 0.76041667 * 192; time = 0.0271s; samplesPerSecond = 7084.6
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.66023382 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0283s; samplesPerSecond = 6787.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.62838263 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0267s; samplesPerSecond = 7183.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.36560063 * 192; EvalErrorPrediction = 0.83854167 * 192; time = 0.0255s; samplesPerSecond = 7523.5
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.50734757 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0271s; samplesPerSecond = 7081.7
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47359624 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0258s; samplesPerSecond = 7435.5
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.55650851 * 192; EvalErrorPrediction = 0.81250000 * 192; time = 0.0262s; samplesPerSecond = 7318.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.13%]: CrossEntropyWithSoftmax = 3.44003656 * 192; EvalErrorPrediction = 0.80208333 * 192; time = 0.0258s; samplesPerSecond = 7451.7
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.35749237 * 192; EvalErrorPrediction = 0.77083333 * 192; time = 0.0253s; samplesPerSecond = 7589.8
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55640565 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0251s; samplesPerSecond = 7660.7
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.32958052 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0260s; samplesPerSecond = 7377.0
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.50833357 * 192; EvalErrorPrediction = 0.84375000 * 192; time = 0.0271s; samplesPerSecond = 7090.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.54460465 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0272s; samplesPerSecond = 7051.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.47456552 * 192; EvalErrorPrediction = 0.81250000 * 192; time = 0.0259s; samplesPerSecond = 7409.7
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.31118826 * 192; EvalErrorPrediction = 0.79687500 * 192; time = 0.0267s; samplesPerSecond = 7179.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.63%]: CrossEntropyWithSoftmax = 3.23043001 * 192; EvalErrorPrediction = 0.78645833 * 192; time = 0.0257s; samplesPerSecond = 7473.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.18504533 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0250s; samplesPerSecond = 7674.8
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23396798 * 192; EvalErrorPrediction = 0.74479167 * 192; time = 0.0265s; samplesPerSecond = 7236.8
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.51781736 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0258s; samplesPerSecond = 7440.1
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 2.94438285 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0249s; samplesPerSecond = 7704.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.05991271 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0248s; samplesPerSecond = 7738.8
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.02642719 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0257s; samplesPerSecond = 7479.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.13208851 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0254s; samplesPerSecond = 7564.7
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.13%]: CrossEntropyWithSoftmax = 2.72670959 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0261s; samplesPerSecond = 7351.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.12385008 * 192; EvalErrorPrediction = 0.80208333 * 192; time = 0.0265s; samplesPerSecond = 7253.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 2.98203627 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0273s; samplesPerSecond = 7045.9
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 2.97060074 * 192; EvalErrorPrediction = 0.75520833 * 192; time = 0.0275s; samplesPerSecond = 6992.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.74938235 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0256s; samplesPerSecond = 7503.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.02481319 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0267s; samplesPerSecond = 7196.9
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.11357602 * 192; EvalErrorPrediction = 0.76041667 * 192; time = 0.0262s; samplesPerSecond = 7327.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.07046118 * 192; EvalErrorPrediction = 0.74479167 * 192; time = 0.0274s; samplesPerSecond = 7019.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.63%]: CrossEntropyWithSoftmax = 2.80324721 * 192; EvalErrorPrediction = 0.72395833 * 192; time = 0.0254s; samplesPerSecond = 7567.1
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.74425310 * 192; EvalErrorPrediction = 0.68750000 * 192; time = 0.0262s; samplesPerSecond = 7336.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 2.95508757 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0265s; samplesPerSecond = 7236.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69285817 * 192; EvalErrorPrediction = 0.67708333 * 192; time = 0.0266s; samplesPerSecond = 7226.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.61222587 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0273s; samplesPerSecond = 7022.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.65528671 * 192; EvalErrorPrediction = 0.65625000 * 192; time = 0.0276s; samplesPerSecond = 6953.0
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.68470586 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0265s; samplesPerSecond = 7234.6
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.78009167 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0279s; samplesPerSecond = 6873.3
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.75177433 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0267s; samplesPerSecond = 7179.2
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.71744614 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0275s; samplesPerSecond = 6991.5
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.78208591 * 192; EvalErrorPrediction = 0.67708333 * 192; time = 0.0259s; samplesPerSecond = 7421.7
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.50297451 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0262s; samplesPerSecond = 7336.4
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97786137 * 192; EvalErrorPrediction = 0.75520833 * 192; time = 0.0249s; samplesPerSecond = 7698.5
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.60160999 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0252s; samplesPerSecond = 7612.1
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.49883465 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0256s; samplesPerSecond = 7511.7
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.58102011 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0264s; samplesPerSecond = 7274.1
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.63%]: CrossEntropyWithSoftmax = 2.55279515 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0278s; samplesPerSecond = 6909.2
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.55386415 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0267s; samplesPerSecond = 7203.2
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.44782705 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0268s; samplesPerSecond = 7170.1
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.71017278 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0284s; samplesPerSecond = 6762.9
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.48043969 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0260s; samplesPerSecond = 7380.4
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.41207736 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0268s; samplesPerSecond = 7170.1
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.71955135 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0251s; samplesPerSecond = 7661.0
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.52658962 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0272s; samplesPerSecond = 7052.6
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.13%]: CrossEntropyWithSoftmax = 2.95080890 * 192; EvalErrorPrediction = 0.77083333 * 192; time = 0.0267s; samplesPerSecond = 7194.0
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.32160400 * 192; EvalErrorPrediction = 0.57812500 * 192; time = 0.0267s; samplesPerSecond = 7185.9
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.21126458 * 192; EvalErrorPrediction = 0.61979167 * 192; time = 0.0282s; samplesPerSecond = 6805.9
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.28277798 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0299s; samplesPerSecond = 6431.3
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.66166449 * 192; EvalErrorPrediction = 0.68750000 * 192; time = 0.0257s; samplesPerSecond = 7461.5
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.41115800 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0251s; samplesPerSecond = 7662.2
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.47460061 * 192; EvalErrorPrediction = 0.65625000 * 192; time = 0.0270s; samplesPerSecond = 7121.9
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.35103706 * 192; EvalErrorPrediction = 0.61979167 * 192; time = 0.0246s; samplesPerSecond = 7805.8
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.63%]: CrossEntropyWithSoftmax = 2.33610739 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0257s; samplesPerSecond = 7457.5
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.29476428 * 192; EvalErrorPrediction = 0.62500000 * 192; time = 0.0256s; samplesPerSecond = 7495.9
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.25023789 * 192; EvalErrorPrediction = 0.59375000 * 192; time = 0.0259s; samplesPerSecond = 7403.7
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.03832213 * 192; EvalErrorPrediction = 0.51041667 * 192; time = 0.0260s; samplesPerSecond = 7379.5
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.37205200 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0253s; samplesPerSecond = 7594.0
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.50009905 * 192; EvalErrorPrediction = 0.65625000 * 192; time = 0.0258s; samplesPerSecond = 7436.4
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.32391497 * 192; EvalErrorPrediction = 0.61979167 * 192; time = 0.0256s; samplesPerSecond = 7506.2
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.56896275 * 192; EvalErrorPrediction = 0.69270833 * 192; time = 0.0256s; samplesPerSecond = 7505.6
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.13%]: CrossEntropyWithSoftmax = 2.44517270 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0274s; samplesPerSecond = 7009.3
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.26122471 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0274s; samplesPerSecond = 7014.0
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.17377959 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0255s; samplesPerSecond = 7537.7
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.22791977 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0259s; samplesPerSecond = 7419.1
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.07889758 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0257s; samplesPerSecond = 7465.3
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.27676510 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0256s; samplesPerSecond = 7508.2
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26006065 * 192; EvalErrorPrediction = 0.60937500 * 192; time = 0.0253s; samplesPerSecond = 7592.5
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.11239689 * 192; EvalErrorPrediction = 0.56250000 * 192; time = 0.0257s; samplesPerSecond = 7459.8
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.63%]: CrossEntropyWithSoftmax = 2.24244492 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0266s; samplesPerSecond = 7206.7
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.14233722 * 192; EvalErrorPrediction = 0.54687500 * 192; time = 0.0272s; samplesPerSecond = 7065.6
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.40359040 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0266s; samplesPerSecond = 7216.1
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.14246064 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0259s; samplesPerSecond = 7418.0
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.18541632 * 192; EvalErrorPrediction = 0.57812500 * 192; time = 0.0264s; samplesPerSecond = 7281.8
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.89472s
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.66660459 * 192; EvalClassificationError = 0.91666667 * 192; time = 0.0454s; samplesPerSecond = 4230.0
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.37978218 * 192; EvalClassificationError = 0.88541667 * 192; time = 0.0254s; samplesPerSecond = 7556.7
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.69675309 * 192; EvalClassificationError = 0.93229167 * 192; time = 0.0265s; samplesPerSecond = 7242.5
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.23088190 * 192; EvalClassificationError = 0.91666667 * 192; time = 0.0262s; samplesPerSecond = 7324.9
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.41262185 * 192; EvalClassificationError = 0.90625000 * 192; time = 0.0250s; samplesPerSecond = 7675.1
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.63%]: CrossEntropyWithSoftmax = 4.35726114 * 192; EvalClassificationError = 0.94270833 * 192; time = 0.0254s; samplesPerSecond = 7551.9
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.22315242 * 192; EvalClassificationError = 0.96354167 * 192; time = 0.0256s; samplesPerSecond = 7509.4
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06862554 * 192; EvalClassificationError = 0.89062500 * 192; time = 0.0258s; samplesPerSecond = 7441.6
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97387863 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0258s; samplesPerSecond = 7440.4
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83243560 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0275s; samplesPerSecond = 6988.9
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79996973 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0267s; samplesPerSecond = 7192.9
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.74380828 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0258s; samplesPerSecond = 7430.9
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73812888 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0248s; samplesPerSecond = 7741.3
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.13%]: CrossEntropyWithSoftmax = 3.69163791 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0254s; samplesPerSecond = 7548.1
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.88173262 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0261s; samplesPerSecond = 7370.4
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.77142322 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0252s; samplesPerSecond = 7610.0
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81063177 * 192; EvalClassificationError = 0.88020833 * 192; time = 0.0263s; samplesPerSecond = 7302.0
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.91374076 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0250s; samplesPerSecond = 7676.0
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.55613122 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0247s; samplesPerSecond = 7758.8
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.64148239 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0270s; samplesPerSecond = 7104.0
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.35544611 * 192; EvalClassificationError = 0.75520833 * 192; time = 0.0263s; samplesPerSecond = 7288.2
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.63%]: CrossEntropyWithSoftmax = 3.40034303 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0247s; samplesPerSecond = 7785.6
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.34712754 * 192; EvalClassificationError = 0.76041667 * 192; time = 0.0271s; samplesPerSecond = 7084.6
|
||||
MPI Rank 1: 08/16/2016 03:19:44: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.66023382 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0283s; samplesPerSecond = 6787.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.62838263 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0267s; samplesPerSecond = 7183.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.36560063 * 192; EvalClassificationError = 0.83854167 * 192; time = 0.0255s; samplesPerSecond = 7523.5
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.50734757 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0271s; samplesPerSecond = 7081.7
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47359624 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0258s; samplesPerSecond = 7435.5
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.55650851 * 192; EvalClassificationError = 0.81250000 * 192; time = 0.0262s; samplesPerSecond = 7318.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.13%]: CrossEntropyWithSoftmax = 3.44003656 * 192; EvalClassificationError = 0.80208333 * 192; time = 0.0258s; samplesPerSecond = 7451.7
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.35749237 * 192; EvalClassificationError = 0.77083333 * 192; time = 0.0253s; samplesPerSecond = 7589.8
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55640565 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0251s; samplesPerSecond = 7660.7
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.32958052 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0260s; samplesPerSecond = 7377.0
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.50833357 * 192; EvalClassificationError = 0.84375000 * 192; time = 0.0271s; samplesPerSecond = 7090.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.54460465 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0272s; samplesPerSecond = 7051.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.47456552 * 192; EvalClassificationError = 0.81250000 * 192; time = 0.0259s; samplesPerSecond = 7409.7
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.31118826 * 192; EvalClassificationError = 0.79687500 * 192; time = 0.0267s; samplesPerSecond = 7179.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.63%]: CrossEntropyWithSoftmax = 3.23043001 * 192; EvalClassificationError = 0.78645833 * 192; time = 0.0257s; samplesPerSecond = 7473.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.18504533 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0250s; samplesPerSecond = 7674.8
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23396798 * 192; EvalClassificationError = 0.74479167 * 192; time = 0.0265s; samplesPerSecond = 7236.8
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.51781736 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0258s; samplesPerSecond = 7440.1
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 2.94438285 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0249s; samplesPerSecond = 7704.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.05991271 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0248s; samplesPerSecond = 7738.8
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.02642719 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0257s; samplesPerSecond = 7479.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.13208851 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0254s; samplesPerSecond = 7564.7
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.13%]: CrossEntropyWithSoftmax = 2.72670959 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0261s; samplesPerSecond = 7351.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.12385008 * 192; EvalClassificationError = 0.80208333 * 192; time = 0.0265s; samplesPerSecond = 7253.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 2.98203627 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0273s; samplesPerSecond = 7045.9
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 2.97060074 * 192; EvalClassificationError = 0.75520833 * 192; time = 0.0275s; samplesPerSecond = 6992.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.74938235 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0256s; samplesPerSecond = 7503.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.02481319 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0267s; samplesPerSecond = 7196.9
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.11357602 * 192; EvalClassificationError = 0.76041667 * 192; time = 0.0262s; samplesPerSecond = 7327.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.07046118 * 192; EvalClassificationError = 0.74479167 * 192; time = 0.0274s; samplesPerSecond = 7019.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.63%]: CrossEntropyWithSoftmax = 2.80324721 * 192; EvalClassificationError = 0.72395833 * 192; time = 0.0254s; samplesPerSecond = 7567.1
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.74425310 * 192; EvalClassificationError = 0.68750000 * 192; time = 0.0262s; samplesPerSecond = 7336.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 2.95508757 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0265s; samplesPerSecond = 7236.3
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69285817 * 192; EvalClassificationError = 0.67708333 * 192; time = 0.0266s; samplesPerSecond = 7226.2
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.61222587 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0273s; samplesPerSecond = 7022.4
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.65528671 * 192; EvalClassificationError = 0.65625000 * 192; time = 0.0276s; samplesPerSecond = 6953.0
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.68470586 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0265s; samplesPerSecond = 7234.6
|
||||
MPI Rank 1: 08/16/2016 03:19:45: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.78009167 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0279s; samplesPerSecond = 6873.3
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.75177433 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0267s; samplesPerSecond = 7179.2
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.71744614 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0275s; samplesPerSecond = 6991.5
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.78208591 * 192; EvalClassificationError = 0.67708333 * 192; time = 0.0259s; samplesPerSecond = 7421.7
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.50297451 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0262s; samplesPerSecond = 7336.4
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97786137 * 192; EvalClassificationError = 0.75520833 * 192; time = 0.0249s; samplesPerSecond = 7698.5
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.60160999 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0252s; samplesPerSecond = 7612.1
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.49883465 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0256s; samplesPerSecond = 7511.7
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.58102011 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0264s; samplesPerSecond = 7274.1
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.63%]: CrossEntropyWithSoftmax = 2.55279515 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0278s; samplesPerSecond = 6909.2
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.55386415 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0267s; samplesPerSecond = 7203.2
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.44782705 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0268s; samplesPerSecond = 7170.1
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.71017278 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0284s; samplesPerSecond = 6762.9
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.48043969 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0260s; samplesPerSecond = 7380.4
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.41207736 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0268s; samplesPerSecond = 7170.1
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.71955135 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0251s; samplesPerSecond = 7661.0
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.52658962 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0272s; samplesPerSecond = 7052.6
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.13%]: CrossEntropyWithSoftmax = 2.95080890 * 192; EvalClassificationError = 0.77083333 * 192; time = 0.0267s; samplesPerSecond = 7194.0
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.32160400 * 192; EvalClassificationError = 0.57812500 * 192; time = 0.0267s; samplesPerSecond = 7185.9
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.21126458 * 192; EvalClassificationError = 0.61979167 * 192; time = 0.0282s; samplesPerSecond = 6805.9
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.28277798 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0299s; samplesPerSecond = 6431.3
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.66166449 * 192; EvalClassificationError = 0.68750000 * 192; time = 0.0257s; samplesPerSecond = 7461.5
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.41115800 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0251s; samplesPerSecond = 7662.2
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.47460061 * 192; EvalClassificationError = 0.65625000 * 192; time = 0.0270s; samplesPerSecond = 7121.9
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.35103706 * 192; EvalClassificationError = 0.61979167 * 192; time = 0.0246s; samplesPerSecond = 7805.8
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.63%]: CrossEntropyWithSoftmax = 2.33610739 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0257s; samplesPerSecond = 7457.5
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.29476428 * 192; EvalClassificationError = 0.62500000 * 192; time = 0.0256s; samplesPerSecond = 7495.9
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.25023789 * 192; EvalClassificationError = 0.59375000 * 192; time = 0.0259s; samplesPerSecond = 7403.7
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.03832213 * 192; EvalClassificationError = 0.51041667 * 192; time = 0.0260s; samplesPerSecond = 7379.5
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.37205200 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0253s; samplesPerSecond = 7594.0
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.50009905 * 192; EvalClassificationError = 0.65625000 * 192; time = 0.0258s; samplesPerSecond = 7436.4
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.32391497 * 192; EvalClassificationError = 0.61979167 * 192; time = 0.0256s; samplesPerSecond = 7506.2
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.56896275 * 192; EvalClassificationError = 0.69270833 * 192; time = 0.0256s; samplesPerSecond = 7505.6
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.13%]: CrossEntropyWithSoftmax = 2.44517270 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0274s; samplesPerSecond = 7009.3
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.26122471 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0274s; samplesPerSecond = 7014.0
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.17377959 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0255s; samplesPerSecond = 7537.7
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.22791977 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0259s; samplesPerSecond = 7419.1
|
||||
MPI Rank 1: 08/16/2016 03:19:46: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.07889758 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0257s; samplesPerSecond = 7465.3
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.27676510 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0256s; samplesPerSecond = 7508.2
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26006065 * 192; EvalClassificationError = 0.60937500 * 192; time = 0.0253s; samplesPerSecond = 7592.5
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.11239689 * 192; EvalClassificationError = 0.56250000 * 192; time = 0.0257s; samplesPerSecond = 7459.8
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.63%]: CrossEntropyWithSoftmax = 2.24244492 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0266s; samplesPerSecond = 7206.7
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.14233722 * 192; EvalClassificationError = 0.54687500 * 192; time = 0.0272s; samplesPerSecond = 7065.6
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.40359040 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0266s; samplesPerSecond = 7216.1
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.14246064 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0259s; samplesPerSecond = 7418.0
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.18541632 * 192; EvalClassificationError = 0.57812500 * 192; time = 0.0264s; samplesPerSecond = 7281.8
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.89472s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.14426480 * 260; EvalErrorPrediction = 0.59615385 * 260; time = 0.0396s; samplesPerSecond = 6564.0
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.29886993 * 276; EvalErrorPrediction = 0.66304348 * 276; time = 0.0352s; samplesPerSecond = 7830.5
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.18974956 * 280; EvalErrorPrediction = 0.65000000 * 280; time = 0.0390s; samplesPerSecond = 7179.3
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.15878861 * 241; EvalErrorPrediction = 0.63900415 * 241; time = 0.0310s; samplesPerSecond = 7773.2
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.02538879 * 295; EvalErrorPrediction = 0.54237288 * 295; time = 0.0404s; samplesPerSecond = 7303.6
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.34684803 * 257; EvalErrorPrediction = 0.61478599 * 257; time = 0.0343s; samplesPerSecond = 7497.1
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.00357501 * 262; EvalErrorPrediction = 0.55725191 * 262; time = 0.0314s; samplesPerSecond = 8338.4
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.14426480 * 260; EvalClassificationError = 0.59615385 * 260; time = 0.0396s; samplesPerSecond = 6564.0
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.29886993 * 276; EvalClassificationError = 0.66304348 * 276; time = 0.0352s; samplesPerSecond = 7830.5
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.18974956 * 280; EvalClassificationError = 0.65000000 * 280; time = 0.0390s; samplesPerSecond = 7179.3
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.15878861 * 241; EvalClassificationError = 0.63900415 * 241; time = 0.0310s; samplesPerSecond = 7773.2
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.02538879 * 295; EvalClassificationError = 0.54237288 * 295; time = 0.0404s; samplesPerSecond = 7303.6
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.34684803 * 257; EvalClassificationError = 0.61478599 * 257; time = 0.0343s; samplesPerSecond = 7497.1
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.00357501 * 262; EvalClassificationError = 0.55725191 * 262; time = 0.0314s; samplesPerSecond = 8338.4
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.31 seconds since last report (0.00 seconds on comm.); 4289 samples processed by 2 workers (2126 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 13.81k samplesPerSecond , throughputPerWorker = 6.91k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.00110195 * 255; EvalErrorPrediction = 0.52941176 * 255; time = 0.0471s; samplesPerSecond = 5416.2
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.04669267 * 279; EvalErrorPrediction = 0.58422939 * 279; time = 0.0359s; samplesPerSecond = 7780.0
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.01079896 * 274; EvalErrorPrediction = 0.54014599 * 274; time = 0.0333s; samplesPerSecond = 8237.6
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 1.91922430 * 269; EvalErrorPrediction = 0.51672862 * 269; time = 0.0352s; samplesPerSecond = 7635.8
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 1.99589323 * 278; EvalErrorPrediction = 0.59712230 * 278; time = 0.0393s; samplesPerSecond = 7075.8
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 2.09043519 * 271; EvalErrorPrediction = 0.54981550 * 271; time = 0.0336s; samplesPerSecond = 8071.7
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.16560966 * 276; EvalErrorPrediction = 0.57608696 * 276; time = 0.0382s; samplesPerSecond = 7224.4
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 2.11114266 * 260; EvalErrorPrediction = 0.56923077 * 260; time = 0.0314s; samplesPerSecond = 8275.8
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.00110195 * 255; EvalClassificationError = 0.52941176 * 255; time = 0.0471s; samplesPerSecond = 5416.2
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.04669267 * 279; EvalClassificationError = 0.58422939 * 279; time = 0.0359s; samplesPerSecond = 7780.0
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.01079896 * 274; EvalClassificationError = 0.54014599 * 274; time = 0.0333s; samplesPerSecond = 8237.6
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 1.91922430 * 269; EvalClassificationError = 0.51672862 * 269; time = 0.0352s; samplesPerSecond = 7635.8
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 1.99589323 * 278; EvalClassificationError = 0.59712230 * 278; time = 0.0393s; samplesPerSecond = 7075.8
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 2.09043519 * 271; EvalClassificationError = 0.54981550 * 271; time = 0.0336s; samplesPerSecond = 8071.7
|
||||
MPI Rank 1: 08/16/2016 03:19:47: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.16560966 * 276; EvalClassificationError = 0.57608696 * 276; time = 0.0382s; samplesPerSecond = 7224.4
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 2.11114266 * 260; EvalClassificationError = 0.56923077 * 260; time = 0.0314s; samplesPerSecond = 8275.8
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.29 seconds since last report (0.00 seconds on comm.); 4253 samples processed by 2 workers (2073 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 14.86k samplesPerSecond , throughputPerWorker = 7.43k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.93313483 * 265; EvalErrorPrediction = 0.53584906 * 265; time = 0.0457s; samplesPerSecond = 5794.6
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.10408761 * 298; EvalErrorPrediction = 0.55704698 * 298; time = 0.0386s; samplesPerSecond = 7712.8
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.93908476 * 274; EvalErrorPrediction = 0.51459854 * 274; time = 0.0348s; samplesPerSecond = 7879.2
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.95290052 * 265; EvalErrorPrediction = 0.53962264 * 265; time = 0.0324s; samplesPerSecond = 8190.6
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.82073681 * 281; EvalErrorPrediction = 0.54804270 * 281; time = 0.0333s; samplesPerSecond = 8432.4
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.92436069 * 252; EvalErrorPrediction = 0.51984127 * 252; time = 0.0343s; samplesPerSecond = 7349.5
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 1.99903754 * 274; EvalErrorPrediction = 0.52554745 * 274; time = 0.0325s; samplesPerSecond = 8442.2
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.81487217 * 258; EvalErrorPrediction = 0.48837209 * 258; time = 0.0314s; samplesPerSecond = 8203.5
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.93313483 * 265; EvalClassificationError = 0.53584906 * 265; time = 0.0457s; samplesPerSecond = 5794.6
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.10408761 * 298; EvalClassificationError = 0.55704698 * 298; time = 0.0386s; samplesPerSecond = 7712.8
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.93908476 * 274; EvalClassificationError = 0.51459854 * 274; time = 0.0348s; samplesPerSecond = 7879.2
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.95290052 * 265; EvalClassificationError = 0.53962264 * 265; time = 0.0324s; samplesPerSecond = 8190.6
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.82073681 * 281; EvalClassificationError = 0.54804270 * 281; time = 0.0333s; samplesPerSecond = 8432.4
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.92436069 * 252; EvalClassificationError = 0.51984127 * 252; time = 0.0343s; samplesPerSecond = 7349.5
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 1.99903754 * 274; EvalClassificationError = 0.52554745 * 274; time = 0.0325s; samplesPerSecond = 8442.2
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.81487217 * 258; EvalClassificationError = 0.48837209 * 258; time = 0.0314s; samplesPerSecond = 8203.5
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.28 seconds since last report (0.00 seconds on comm.); 4246 samples processed by 2 workers (2102 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 15.10k samplesPerSecond , throughputPerWorker = 7.55k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 1.94358599 * 271; EvalErrorPrediction = 0.55350554 * 271; time = 0.0478s; samplesPerSecond = 5666.6
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.96453147 * 278; EvalErrorPrediction = 0.52517986 * 278; time = 0.0354s; samplesPerSecond = 7858.2
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.90278464 * 286; EvalErrorPrediction = 0.54895105 * 286; time = 0.0338s; samplesPerSecond = 8468.8
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.91967982 * 170; EvalErrorPrediction = 0.53529412 * 170; time = 0.0214s; samplesPerSecond = 7957.7
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 1.94358599 * 271; EvalClassificationError = 0.55350554 * 271; time = 0.0478s; samplesPerSecond = 5666.6
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.96453147 * 278; EvalClassificationError = 0.52517986 * 278; time = 0.0354s; samplesPerSecond = 7858.2
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.90278464 * 286; EvalClassificationError = 0.54895105 * 286; time = 0.0338s; samplesPerSecond = 8468.8
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.91967982 * 170; EvalClassificationError = 0.53529412 * 170; time = 0.0214s; samplesPerSecond = 7957.7
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.12-seconds latency this time; accumulated time on sync point = 0.12 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.64 seconds since last report (0.39 seconds on comm.); 7692 samples processed by 2 workers (904 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 12.03k samplesPerSecond , throughputPerWorker = 6.01k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.01405635 * 20480; EvalErrorPrediction = 0.55107422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.51892s
|
||||
MPI Rank 1: 08/16/2016 03:19:48: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.01405635 * 20480; EvalClassificationError = 0.55107422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.51892s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.90600076 * 1130; EvalErrorPrediction = 0.52300885 * 1130; time = 0.1521s; samplesPerSecond = 7431.5
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.90600076 * 1130; EvalClassificationError = 0.52300885 * 1130; time = 0.1521s; samplesPerSecond = 7431.5
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.31 seconds since last report (0.00 seconds on comm.); 4885 samples processed by 2 workers (2293 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 15.81k samplesPerSecond , throughputPerWorker = 7.90k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.85164148 * 1163; EvalErrorPrediction = 0.52708512 * 1163; time = 0.1516s; samplesPerSecond = 7671.7
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.94881175 * 1085; EvalErrorPrediction = 0.52534562 * 1085; time = 0.1121s; samplesPerSecond = 9679.0
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.85164148 * 1163; EvalClassificationError = 0.52708512 * 1163; time = 0.1516s; samplesPerSecond = 7671.7
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.94881175 * 1085; EvalClassificationError = 0.52534562 * 1085; time = 0.1121s; samplesPerSecond = 9679.0
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.03 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.28 seconds since last report (0.00 seconds on comm.); 4826 samples processed by 2 workers (2249 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 17.03k samplesPerSecond , throughputPerWorker = 8.52k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.90020794 * 1164; EvalErrorPrediction = 0.55326460 * 1164; time = 0.1699s; samplesPerSecond = 6850.9
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.98043998 * 1167; EvalErrorPrediction = 0.56640960 * 1167; time = 0.1190s; samplesPerSecond = 9806.4
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.90020794 * 1164; EvalClassificationError = 0.55326460 * 1164; time = 0.1699s; samplesPerSecond = 6850.9
|
||||
MPI Rank 1: 08/16/2016 03:19:49: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.98043998 * 1167; EvalClassificationError = 0.56640960 * 1167; time = 0.1190s; samplesPerSecond = 9806.4
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.07 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.29 seconds since last report (0.00 seconds on comm.); 4903 samples processed by 2 workers (2326 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 17.15k samplesPerSecond , throughputPerWorker = 8.58k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.89797380 * 1159; EvalErrorPrediction = 0.53321829 * 1159; time = 0.1655s; samplesPerSecond = 7004.1
|
||||
MPI Rank 1: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.95982447 * 823; EvalErrorPrediction = 0.54556501 * 823; time = 0.0955s; samplesPerSecond = 8615.9
|
||||
MPI Rank 1: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.89797380 * 1159; EvalClassificationError = 0.53321829 * 1159; time = 0.1655s; samplesPerSecond = 7004.1
|
||||
MPI Rank 1: 08/16/2016 03:19:50: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.95982447 * 823; EvalClassificationError = 0.54556501 * 823; time = 0.0955s; samplesPerSecond = 8615.9
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.16-seconds latency this time; accumulated time on sync point = 0.22 seconds , average latency = 0.06 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.46 seconds since last report (0.19 seconds on comm.); 5866 samples processed by 2 workers (823 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 12.66k samplesPerSecond , throughputPerWorker = 6.33k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:50: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.92634052 * 20480; EvalErrorPrediction = 0.54199219 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.3433s
|
||||
MPI Rank 1: 08/16/2016 03:19:50: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.92634052 * 20480; EvalClassificationError = 0.54199219 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.3433s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:50: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:50: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:19:50: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.86867516 * 1149; EvalErrorPrediction = 0.52654482 * 1149; time = 0.1051s; samplesPerSecond = 10929.6
|
||||
MPI Rank 1: 08/16/2016 03:19:50: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.86867516 * 1149; EvalClassificationError = 0.52654482 * 1149; time = 0.1051s; samplesPerSecond = 10929.6
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.03 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.27 seconds since last report (0.00 seconds on comm.); 4901 samples processed by 2 workers (2351 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 17.88k samplesPerSecond , throughputPerWorker = 8.94k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.91862080 * 1202; EvalErrorPrediction = 0.52995008 * 1202; time = 0.1639s; samplesPerSecond = 7332.5
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.87811845 * 1130; EvalErrorPrediction = 0.51858407 * 1130; time = 0.0986s; samplesPerSecond = 11457.3
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.91862080 * 1202; EvalClassificationError = 0.52995008 * 1202; time = 0.1639s; samplesPerSecond = 7332.5
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.87811845 * 1130; EvalClassificationError = 0.51858407 * 1130; time = 0.0986s; samplesPerSecond = 11457.3
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.04-seconds latency this time; accumulated time on sync point = 0.07 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.27 seconds since last report (0.00 seconds on comm.); 4836 samples processed by 2 workers (2317 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 17.97k samplesPerSecond , throughputPerWorker = 8.98k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.82542386 * 1187; EvalErrorPrediction = 0.50463353 * 1187; time = 0.1691s; samplesPerSecond = 7017.8
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.90363851 * 1202; EvalErrorPrediction = 0.53494176 * 1202; time = 0.1132s; samplesPerSecond = 10614.8
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.82542386 * 1187; EvalClassificationError = 0.50463353 * 1187; time = 0.1691s; samplesPerSecond = 7017.8
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.90363851 * 1202; EvalClassificationError = 0.53494176 * 1202; time = 0.1132s; samplesPerSecond = 10614.8
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.05-seconds latency this time; accumulated time on sync point = 0.12 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.29 seconds since last report (0.00 seconds on comm.); 4952 samples processed by 2 workers (2401 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 17.08k samplesPerSecond , throughputPerWorker = 8.54k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.88586828 * 1199; EvalErrorPrediction = 0.53127606 * 1199; time = 0.1753s; samplesPerSecond = 6839.2
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.88599726 * 817; EvalErrorPrediction = 0.52998776 * 817; time = 0.0685s; samplesPerSecond = 11933.3
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.88586828 * 1199; EvalClassificationError = 0.53127606 * 1199; time = 0.1753s; samplesPerSecond = 6839.2
|
||||
MPI Rank 1: 08/16/2016 03:19:51: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.88599726 * 817; EvalClassificationError = 0.52998776 * 817; time = 0.0685s; samplesPerSecond = 11933.3
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.16-seconds latency this time; accumulated time on sync point = 0.28 seconds , average latency = 0.07 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.42 seconds since last report (0.18 seconds on comm.); 5791 samples processed by 2 workers (817 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 13.82k samplesPerSecond , throughputPerWorker = 6.91k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.87410515 * 20480; EvalErrorPrediction = 0.52202148 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=1.25393s
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.87410515 * 20480; EvalClassificationError = 0.52202148 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=1.25393s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Starting Epoch 5: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 4: frames [81920..102400] (first utterance at frame 81920), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.84239919 * 1209; EvalErrorPrediction = 0.49379653 * 1209; time = 0.1164s; samplesPerSecond = 10382.2
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.84239919 * 1209; EvalClassificationError = 0.49379653 * 1209; time = 0.1164s; samplesPerSecond = 10382.2
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.29 seconds since last report (0.00 seconds on comm.); 4919 samples processed by 2 workers (2426 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 17.01k samplesPerSecond , throughputPerWorker = 8.51k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.90901096 * 1217; EvalErrorPrediction = 0.52095316 * 1217; time = 0.1678s; samplesPerSecond = 7252.6
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.86399282 * 1206; EvalErrorPrediction = 0.50663350 * 1206; time = 0.1037s; samplesPerSecond = 11624.7
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.90901096 * 1217; EvalClassificationError = 0.52095316 * 1217; time = 0.1678s; samplesPerSecond = 7252.6
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.86399282 * 1206; EvalClassificationError = 0.50663350 * 1206; time = 0.1037s; samplesPerSecond = 11624.7
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.05-seconds latency this time; accumulated time on sync point = 0.07 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.27 seconds since last report (0.00 seconds on comm.); 4899 samples processed by 2 workers (2419 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 17.86k samplesPerSecond , throughputPerWorker = 8.93k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.93499820 * 1213; EvalErrorPrediction = 0.53668590 * 1213; time = 0.1692s; samplesPerSecond = 7168.9
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.81107374 * 1147; EvalErrorPrediction = 0.50566696 * 1147; time = 0.1059s; samplesPerSecond = 10827.5
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.93499820 * 1213; EvalClassificationError = 0.53668590 * 1213; time = 0.1692s; samplesPerSecond = 7168.9
|
||||
MPI Rank 1: 08/16/2016 03:19:52: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.81107374 * 1147; EvalClassificationError = 0.50566696 * 1147; time = 0.1059s; samplesPerSecond = 10827.5
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.10 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.25 seconds since last report (0.00 seconds on comm.); 4829 samples processed by 2 workers (2359 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 19.18k samplesPerSecond , throughputPerWorker = 9.59k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.99553623 * 1212; EvalErrorPrediction = 0.55198020 * 1212; time = 0.1444s; samplesPerSecond = 8394.1
|
||||
MPI Rank 1: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.85509924 * 809; EvalErrorPrediction = 0.51050680 * 809; time = 0.0682s; samplesPerSecond = 11867.4
|
||||
MPI Rank 1: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.99553623 * 1212; EvalClassificationError = 0.55198020 * 1212; time = 0.1444s; samplesPerSecond = 8394.1
|
||||
MPI Rank 1: 08/16/2016 03:19:53: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.85509924 * 809; EvalClassificationError = 0.51050680 * 809; time = 0.0682s; samplesPerSecond = 11867.4
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.16-seconds latency this time; accumulated time on sync point = 0.25 seconds , average latency = 0.06 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.41 seconds since last report (0.17 seconds on comm.); 5833 samples processed by 2 workers (809 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 14.06k samplesPerSecond , throughputPerWorker = 7.03k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:19:53: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.85694655 * 20480; EvalErrorPrediction = 0.51362305 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-005; epochTime=1.23172s
|
||||
MPI Rank 1: 08/16/2016 03:19:53: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.85694655 * 20480; EvalClassificationError = 0.51362305 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-005; epochTime=1.23172s
|
||||
MPI Rank 1: 08/16/2016 03:19:53: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:19:53: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -104,7 +104,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -130,7 +130,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -212,7 +212,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -238,7 +238,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -326,7 +326,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -352,7 +352,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -452,7 +452,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -481,7 +481,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -504,7 +504,7 @@ MPI Rank 0: 08/16/2016 03:20:03: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:20:03: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:03: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:20:03: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:20:03: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -557,113 +557,113 @@ MPI Rank 0: 08/16/2016 03:20:09: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.68332137 * 192; EvalErrorPrediction = 0.98958333 * 192; time = 0.0205s; samplesPerSecond = 9370.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.42217834 * 192; EvalErrorPrediction = 0.89062500 * 192; time = 0.0205s; samplesPerSecond = 9375.9
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.78361173 * 192; EvalErrorPrediction = 0.93229167 * 192; time = 0.0216s; samplesPerSecond = 8886.8
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.28266373 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0220s; samplesPerSecond = 8722.9
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.48941255 * 192; EvalErrorPrediction = 0.89583333 * 192; time = 0.0217s; samplesPerSecond = 8849.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.63%]: CrossEntropyWithSoftmax = 4.39312485 * 192; EvalErrorPrediction = 0.94270833 * 192; time = 0.0217s; samplesPerSecond = 8867.5
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.26351131 * 192; EvalErrorPrediction = 0.98437500 * 192; time = 0.0218s; samplesPerSecond = 8791.2
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06459300 * 192; EvalErrorPrediction = 0.90625000 * 192; time = 0.0216s; samplesPerSecond = 8878.2
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97918086 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0220s; samplesPerSecond = 8746.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83987406 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0205s; samplesPerSecond = 9363.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79421365 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0206s; samplesPerSecond = 9334.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.76043875 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0206s; samplesPerSecond = 9341.7
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73154819 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0205s; samplesPerSecond = 9357.6
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.13%]: CrossEntropyWithSoftmax = 3.68435113 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0206s; samplesPerSecond = 9339.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89304289 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0205s; samplesPerSecond = 9349.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.76826980 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0205s; samplesPerSecond = 9349.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81256010 * 192; EvalErrorPrediction = 0.91145833 * 192; time = 0.0207s; samplesPerSecond = 9287.9
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.92133036 * 192; EvalErrorPrediction = 0.87500000 * 192; time = 0.0205s; samplesPerSecond = 9357.2
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56128851 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0205s; samplesPerSecond = 9346.7
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.67830123 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0205s; samplesPerSecond = 9370.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.36612487 * 192; EvalErrorPrediction = 0.76041667 * 192; time = 0.0204s; samplesPerSecond = 9392.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.63%]: CrossEntropyWithSoftmax = 3.41785793 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9351.3
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.37660990 * 192; EvalErrorPrediction = 0.79166667 * 192; time = 0.0205s; samplesPerSecond = 9358.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.68727319 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0205s; samplesPerSecond = 9371.3
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.64994161 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0204s; samplesPerSecond = 9397.5
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.37700933 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0205s; samplesPerSecond = 9352.2
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51711918 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0204s; samplesPerSecond = 9393.3
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47828208 * 192; EvalErrorPrediction = 0.80208333 * 192; time = 0.0205s; samplesPerSecond = 9368.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.54864536 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0206s; samplesPerSecond = 9338.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.13%]: CrossEntropyWithSoftmax = 3.43454656 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9353.5
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.36875092 * 192; EvalErrorPrediction = 0.79687500 * 192; time = 0.0205s; samplesPerSecond = 9359.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55401747 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0205s; samplesPerSecond = 9349.9
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.34809828 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9354.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.49450178 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0206s; samplesPerSecond = 9328.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.55445642 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0205s; samplesPerSecond = 9365.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.49293717 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9361.3
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.32494958 * 192; EvalErrorPrediction = 0.79687500 * 192; time = 0.0205s; samplesPerSecond = 9372.7
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.63%]: CrossEntropyWithSoftmax = 3.28851645 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0206s; samplesPerSecond = 9310.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.19411841 * 192; EvalErrorPrediction = 0.78125000 * 192; time = 0.0206s; samplesPerSecond = 9310.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.25028950 * 192; EvalErrorPrediction = 0.78125000 * 192; time = 0.0206s; samplesPerSecond = 9329.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.53445091 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0204s; samplesPerSecond = 9407.2
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00326214 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0206s; samplesPerSecond = 9327.6
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.10787619 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0205s; samplesPerSecond = 9354.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.05588285 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0206s; samplesPerSecond = 9331.3
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18197888 * 192; EvalErrorPrediction = 0.74479167 * 192; time = 0.0205s; samplesPerSecond = 9343.5
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.13%]: CrossEntropyWithSoftmax = 2.78138232 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0205s; samplesPerSecond = 9349.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17441293 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0204s; samplesPerSecond = 9388.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03537426 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0205s; samplesPerSecond = 9347.6
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.00595982 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0205s; samplesPerSecond = 9346.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79115323 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9358.5
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.04097805 * 192; EvalErrorPrediction = 0.69791667 * 192; time = 0.0206s; samplesPerSecond = 9339.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.15935323 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0205s; samplesPerSecond = 9351.3
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11947251 * 192; EvalErrorPrediction = 0.75000000 * 192; time = 0.0206s; samplesPerSecond = 9338.1
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.63%]: CrossEntropyWithSoftmax = 2.86232384 * 192; EvalErrorPrediction = 0.72395833 * 192; time = 0.0206s; samplesPerSecond = 9341.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.77742114 * 192; EvalErrorPrediction = 0.70312500 * 192; time = 0.0206s; samplesPerSecond = 9339.4
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00411374 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0205s; samplesPerSecond = 9344.4
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69740857 * 192; EvalErrorPrediction = 0.68750000 * 192; time = 0.0204s; samplesPerSecond = 9409.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64948203 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0206s; samplesPerSecond = 9338.1
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71417618 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0205s; samplesPerSecond = 9383.7
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.74281938 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0206s; samplesPerSecond = 9328.1
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.81346277 * 192; EvalErrorPrediction = 0.72916667 * 192; time = 0.0205s; samplesPerSecond = 9384.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79862588 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0206s; samplesPerSecond = 9318.6
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.76655026 * 192; EvalErrorPrediction = 0.71354167 * 192; time = 0.0205s; samplesPerSecond = 9344.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.75908943 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0206s; samplesPerSecond = 9340.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.53548199 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0204s; samplesPerSecond = 9411.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97589950 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0206s; samplesPerSecond = 9329.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.64996722 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0205s; samplesPerSecond = 9349.4
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52128374 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9345.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.64228785 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0205s; samplesPerSecond = 9350.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.63%]: CrossEntropyWithSoftmax = 2.57199182 * 192; EvalErrorPrediction = 0.67708333 * 192; time = 0.0205s; samplesPerSecond = 9344.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.58100902 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0205s; samplesPerSecond = 9347.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.48555431 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0206s; samplesPerSecond = 9334.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.75336137 * 192; EvalErrorPrediction = 0.67187500 * 192; time = 0.0206s; samplesPerSecond = 9340.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.49193178 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0205s; samplesPerSecond = 9352.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.46098647 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0206s; samplesPerSecond = 9341.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.74322101 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0205s; samplesPerSecond = 9351.7
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.55837089 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0205s; samplesPerSecond = 9344.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.13%]: CrossEntropyWithSoftmax = 2.98288510 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0206s; samplesPerSecond = 9329.4
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.36667287 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0204s; samplesPerSecond = 9391.5
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.25169614 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0205s; samplesPerSecond = 9364.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.31564120 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0206s; samplesPerSecond = 9323.6
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.70894816 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0205s; samplesPerSecond = 9351.3
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.44991146 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0206s; samplesPerSecond = 9322.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.51856232 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0206s; samplesPerSecond = 9331.3
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.38498228 * 192; EvalErrorPrediction = 0.61979167 * 192; time = 0.0205s; samplesPerSecond = 9347.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.63%]: CrossEntropyWithSoftmax = 2.38080698 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0205s; samplesPerSecond = 9351.7
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.34294967 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9352.6
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.30340167 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0206s; samplesPerSecond = 9324.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.08323277 * 192; EvalErrorPrediction = 0.52604167 * 192; time = 0.0205s; samplesPerSecond = 9359.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.43589953 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0206s; samplesPerSecond = 9329.9
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.53399486 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0206s; samplesPerSecond = 9340.8
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33995708 * 192; EvalErrorPrediction = 0.60416667 * 192; time = 0.0205s; samplesPerSecond = 9347.2
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.62970864 * 192; EvalErrorPrediction = 0.68229167 * 192; time = 0.0206s; samplesPerSecond = 9329.0
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.13%]: CrossEntropyWithSoftmax = 2.47993989 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0206s; samplesPerSecond = 9337.6
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30935254 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0206s; samplesPerSecond = 9340.3
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.22022265 * 192; EvalErrorPrediction = 0.59375000 * 192; time = 0.0205s; samplesPerSecond = 9347.6
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.28060247 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0205s; samplesPerSecond = 9367.7
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.13349744 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0206s; samplesPerSecond = 9312.7
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.29751600 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0204s; samplesPerSecond = 9405.3
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.29319143 * 192; EvalErrorPrediction = 0.60416667 * 192; time = 0.0206s; samplesPerSecond = 9342.6
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.14551002 * 192; EvalErrorPrediction = 0.55729167 * 192; time = 0.0204s; samplesPerSecond = 9401.2
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.63%]: CrossEntropyWithSoftmax = 2.26930674 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0206s; samplesPerSecond = 9329.9
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.17383355 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0204s; samplesPerSecond = 9399.8
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.43111882 * 192; EvalErrorPrediction = 0.67187500 * 192; time = 0.0205s; samplesPerSecond = 9356.3
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.18011227 * 192; EvalErrorPrediction = 0.59895833 * 192; time = 0.0206s; samplesPerSecond = 9338.5
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.21682707 * 192; EvalErrorPrediction = 0.56250000 * 192; time = 0.0205s; samplesPerSecond = 9361.3
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.24705s
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.68332137 * 192; EvalClassificationError = 0.98958333 * 192; time = 0.0205s; samplesPerSecond = 9370.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.42217834 * 192; EvalClassificationError = 0.89062500 * 192; time = 0.0205s; samplesPerSecond = 9375.9
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.78361173 * 192; EvalClassificationError = 0.93229167 * 192; time = 0.0216s; samplesPerSecond = 8886.8
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.28266373 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0220s; samplesPerSecond = 8722.9
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.48941255 * 192; EvalClassificationError = 0.89583333 * 192; time = 0.0217s; samplesPerSecond = 8849.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.63%]: CrossEntropyWithSoftmax = 4.39312485 * 192; EvalClassificationError = 0.94270833 * 192; time = 0.0217s; samplesPerSecond = 8867.5
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.26351131 * 192; EvalClassificationError = 0.98437500 * 192; time = 0.0218s; samplesPerSecond = 8791.2
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06459300 * 192; EvalClassificationError = 0.90625000 * 192; time = 0.0216s; samplesPerSecond = 8878.2
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97918086 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0220s; samplesPerSecond = 8746.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83987406 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0205s; samplesPerSecond = 9363.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79421365 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0206s; samplesPerSecond = 9334.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.76043875 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0206s; samplesPerSecond = 9341.7
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73154819 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0205s; samplesPerSecond = 9357.6
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.13%]: CrossEntropyWithSoftmax = 3.68435113 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0206s; samplesPerSecond = 9339.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89304289 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0205s; samplesPerSecond = 9349.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.76826980 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0205s; samplesPerSecond = 9349.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81256010 * 192; EvalClassificationError = 0.91145833 * 192; time = 0.0207s; samplesPerSecond = 9287.9
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.92133036 * 192; EvalClassificationError = 0.87500000 * 192; time = 0.0205s; samplesPerSecond = 9357.2
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56128851 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0205s; samplesPerSecond = 9346.7
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.67830123 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0205s; samplesPerSecond = 9370.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.36612487 * 192; EvalClassificationError = 0.76041667 * 192; time = 0.0204s; samplesPerSecond = 9392.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.63%]: CrossEntropyWithSoftmax = 3.41785793 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9351.3
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.37660990 * 192; EvalClassificationError = 0.79166667 * 192; time = 0.0205s; samplesPerSecond = 9358.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.68727319 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0205s; samplesPerSecond = 9371.3
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.64994161 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0204s; samplesPerSecond = 9397.5
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.37700933 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0205s; samplesPerSecond = 9352.2
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51711918 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0204s; samplesPerSecond = 9393.3
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47828208 * 192; EvalClassificationError = 0.80208333 * 192; time = 0.0205s; samplesPerSecond = 9368.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.54864536 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0206s; samplesPerSecond = 9338.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.13%]: CrossEntropyWithSoftmax = 3.43454656 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9353.5
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.36875092 * 192; EvalClassificationError = 0.79687500 * 192; time = 0.0205s; samplesPerSecond = 9359.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55401747 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0205s; samplesPerSecond = 9349.9
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.34809828 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9354.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.49450178 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0206s; samplesPerSecond = 9328.1
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.55445642 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0205s; samplesPerSecond = 9365.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.49293717 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9361.3
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.32494958 * 192; EvalClassificationError = 0.79687500 * 192; time = 0.0205s; samplesPerSecond = 9372.7
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.63%]: CrossEntropyWithSoftmax = 3.28851645 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0206s; samplesPerSecond = 9310.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.19411841 * 192; EvalClassificationError = 0.78125000 * 192; time = 0.0206s; samplesPerSecond = 9310.0
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.25028950 * 192; EvalClassificationError = 0.78125000 * 192; time = 0.0206s; samplesPerSecond = 9329.4
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.53445091 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0204s; samplesPerSecond = 9407.2
|
||||
MPI Rank 0: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00326214 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0206s; samplesPerSecond = 9327.6
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.10787619 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0205s; samplesPerSecond = 9354.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.05588285 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0206s; samplesPerSecond = 9331.3
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18197888 * 192; EvalClassificationError = 0.74479167 * 192; time = 0.0205s; samplesPerSecond = 9343.5
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.13%]: CrossEntropyWithSoftmax = 2.78138232 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0205s; samplesPerSecond = 9349.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17441293 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0204s; samplesPerSecond = 9388.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03537426 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0205s; samplesPerSecond = 9347.6
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.00595982 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0205s; samplesPerSecond = 9346.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79115323 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9358.5
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.04097805 * 192; EvalClassificationError = 0.69791667 * 192; time = 0.0206s; samplesPerSecond = 9339.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.15935323 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0205s; samplesPerSecond = 9351.3
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11947251 * 192; EvalClassificationError = 0.75000000 * 192; time = 0.0206s; samplesPerSecond = 9338.1
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.63%]: CrossEntropyWithSoftmax = 2.86232384 * 192; EvalClassificationError = 0.72395833 * 192; time = 0.0206s; samplesPerSecond = 9341.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.77742114 * 192; EvalClassificationError = 0.70312500 * 192; time = 0.0206s; samplesPerSecond = 9339.4
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00411374 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0205s; samplesPerSecond = 9344.4
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69740857 * 192; EvalClassificationError = 0.68750000 * 192; time = 0.0204s; samplesPerSecond = 9409.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64948203 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0206s; samplesPerSecond = 9338.1
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71417618 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0205s; samplesPerSecond = 9383.7
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.74281938 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0206s; samplesPerSecond = 9328.1
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.81346277 * 192; EvalClassificationError = 0.72916667 * 192; time = 0.0205s; samplesPerSecond = 9384.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79862588 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0206s; samplesPerSecond = 9318.6
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.76655026 * 192; EvalClassificationError = 0.71354167 * 192; time = 0.0205s; samplesPerSecond = 9344.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.75908943 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0206s; samplesPerSecond = 9340.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.53548199 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0204s; samplesPerSecond = 9411.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97589950 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0206s; samplesPerSecond = 9329.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.64996722 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0205s; samplesPerSecond = 9349.4
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52128374 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9345.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.64228785 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0205s; samplesPerSecond = 9350.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.63%]: CrossEntropyWithSoftmax = 2.57199182 * 192; EvalClassificationError = 0.67708333 * 192; time = 0.0205s; samplesPerSecond = 9344.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.58100902 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0205s; samplesPerSecond = 9347.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.48555431 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0206s; samplesPerSecond = 9334.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.75336137 * 192; EvalClassificationError = 0.67187500 * 192; time = 0.0206s; samplesPerSecond = 9340.8
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.49193178 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0205s; samplesPerSecond = 9352.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.46098647 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0206s; samplesPerSecond = 9341.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.74322101 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0205s; samplesPerSecond = 9351.7
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.55837089 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0205s; samplesPerSecond = 9344.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.13%]: CrossEntropyWithSoftmax = 2.98288510 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0206s; samplesPerSecond = 9329.4
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.36667287 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0204s; samplesPerSecond = 9391.5
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.25169614 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0205s; samplesPerSecond = 9364.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.31564120 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0206s; samplesPerSecond = 9323.6
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.70894816 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0205s; samplesPerSecond = 9351.3
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.44991146 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0206s; samplesPerSecond = 9322.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.51856232 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0206s; samplesPerSecond = 9331.3
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.38498228 * 192; EvalClassificationError = 0.61979167 * 192; time = 0.0205s; samplesPerSecond = 9347.2
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.63%]: CrossEntropyWithSoftmax = 2.38080698 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0205s; samplesPerSecond = 9351.7
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.34294967 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9352.6
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.30340167 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0206s; samplesPerSecond = 9324.0
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.08323277 * 192; EvalClassificationError = 0.52604167 * 192; time = 0.0205s; samplesPerSecond = 9359.9
|
||||
MPI Rank 0: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.43589953 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0206s; samplesPerSecond = 9329.9
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.53399486 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0206s; samplesPerSecond = 9340.8
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33995708 * 192; EvalClassificationError = 0.60416667 * 192; time = 0.0205s; samplesPerSecond = 9347.2
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.62970864 * 192; EvalClassificationError = 0.68229167 * 192; time = 0.0206s; samplesPerSecond = 9329.0
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.13%]: CrossEntropyWithSoftmax = 2.47993989 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0206s; samplesPerSecond = 9337.6
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30935254 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0206s; samplesPerSecond = 9340.3
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.22022265 * 192; EvalClassificationError = 0.59375000 * 192; time = 0.0205s; samplesPerSecond = 9347.6
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.28060247 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0205s; samplesPerSecond = 9367.7
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.13349744 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0206s; samplesPerSecond = 9312.7
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.29751600 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0204s; samplesPerSecond = 9405.3
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.29319143 * 192; EvalClassificationError = 0.60416667 * 192; time = 0.0206s; samplesPerSecond = 9342.6
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.14551002 * 192; EvalClassificationError = 0.55729167 * 192; time = 0.0204s; samplesPerSecond = 9401.2
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.63%]: CrossEntropyWithSoftmax = 2.26930674 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0206s; samplesPerSecond = 9329.9
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.17383355 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0204s; samplesPerSecond = 9399.8
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.43111882 * 192; EvalClassificationError = 0.67187500 * 192; time = 0.0205s; samplesPerSecond = 9356.3
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.18011227 * 192; EvalClassificationError = 0.59895833 * 192; time = 0.0206s; samplesPerSecond = 9338.5
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.21682707 * 192; EvalClassificationError = 0.56250000 * 192; time = 0.0205s; samplesPerSecond = 9361.3
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.24705s
|
||||
MPI Rank 0: 08/16/2016 03:20:11: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBM@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -671,46 +671,46 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.20431953 * 508; EvalErrorPrediction = 0.59842520 * 508; time = 0.0273s; samplesPerSecond = 18577.4
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.19237836 * 492; EvalErrorPrediction = 0.57723577 * 492; time = 0.0198s; samplesPerSecond = 24866.1
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.17526222 * 488; EvalErrorPrediction = 0.59836066 * 488; time = 0.0224s; samplesPerSecond = 21761.4
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.23483521 * 527; EvalErrorPrediction = 0.59582543 * 527; time = 0.0441s; samplesPerSecond = 11942.0
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.20431953 * 508; EvalClassificationError = 0.59842520 * 508; time = 0.0273s; samplesPerSecond = 18577.4
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.19237836 * 492; EvalClassificationError = 0.57723577 * 492; time = 0.0198s; samplesPerSecond = 24866.1
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.17526222 * 488; EvalClassificationError = 0.59836066 * 488; time = 0.0224s; samplesPerSecond = 21761.4
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.23483521 * 527; EvalClassificationError = 0.59582543 * 527; time = 0.0441s; samplesPerSecond = 11942.0
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.07-seconds latency this time; accumulated time on sync point = 0.07 seconds , average latency = 0.07 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.24 seconds since last report (0.01 seconds on comm.); 4289 samples processed by 2 workers (2163 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 18.06k samplesPerSecond , throughputPerWorker = 9.03k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.00528284 * 473; EvalErrorPrediction = 0.53911205 * 473; time = 0.1345s; samplesPerSecond = 3516.6
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.08558038 * 511; EvalErrorPrediction = 0.54990215 * 511; time = 0.0267s; samplesPerSecond = 19109.2
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.08506441 * 506; EvalErrorPrediction = 0.54940711 * 506; time = 0.0264s; samplesPerSecond = 19198.7
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.12168501 * 513; EvalErrorPrediction = 0.52241715 * 513; time = 0.0268s; samplesPerSecond = 19165.4
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.00528284 * 473; EvalClassificationError = 0.53911205 * 473; time = 0.1345s; samplesPerSecond = 3516.6
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.08558038 * 511; EvalClassificationError = 0.54990215 * 511; time = 0.0267s; samplesPerSecond = 19109.2
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.08506441 * 506; EvalClassificationError = 0.54940711 * 506; time = 0.0264s; samplesPerSecond = 19198.7
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.12168501 * 513; EvalClassificationError = 0.52241715 * 513; time = 0.0268s; samplesPerSecond = 19165.4
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.05-seconds latency this time; accumulated time on sync point = 0.12 seconds , average latency = 0.06 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.18 seconds since last report (0.01 seconds on comm.); 4253 samples processed by 2 workers (2180 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 23.54k samplesPerSecond , throughputPerWorker = 11.77k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.08058681 * 489; EvalErrorPrediction = 0.56646217 * 489; time = 0.0896s; samplesPerSecond = 5458.5
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07411044 * 494; EvalErrorPrediction = 0.55060729 * 494; time = 0.0275s; samplesPerSecond = 17987.8
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 2.12310728 * 499; EvalErrorPrediction = 0.58316633 * 499; time = 0.0270s; samplesPerSecond = 18452.8
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.06918046 * 490; EvalErrorPrediction = 0.57142857 * 490; time = 0.0263s; samplesPerSecond = 18658.1
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.08058681 * 489; EvalClassificationError = 0.56646217 * 489; time = 0.0896s; samplesPerSecond = 5458.5
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07411044 * 494; EvalClassificationError = 0.55060729 * 494; time = 0.0275s; samplesPerSecond = 17987.8
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 2.12310728 * 499; EvalClassificationError = 0.58316633 * 499; time = 0.0270s; samplesPerSecond = 18452.8
|
||||
MPI Rank 0: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.06918046 * 490; EvalClassificationError = 0.57142857 * 490; time = 0.0263s; samplesPerSecond = 18658.1
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.15 seconds , average latency = 0.05 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.19 seconds since last report (0.00 seconds on comm.); 4246 samples processed by 2 workers (2144 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 22.72k samplesPerSecond , throughputPerWorker = 11.36k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 1.93293190 * 497; EvalErrorPrediction = 0.50905433 * 497; time = 0.0989s; samplesPerSecond = 5024.7
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.13718669 * 492; EvalErrorPrediction = 0.59552846 * 492; time = 0.0262s; samplesPerSecond = 18811.7
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 1.91004653 * 508; EvalErrorPrediction = 0.54527559 * 508; time = 0.0266s; samplesPerSecond = 19105.6
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.97341931 * 503; EvalErrorPrediction = 0.52286282 * 503; time = 0.0264s; samplesPerSecond = 19054.5
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.07837958 * 470; EvalErrorPrediction = 0.57021277 * 470; time = 0.0165s; samplesPerSecond = 28431.4
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.91466321 * 494; EvalErrorPrediction = 0.54858300 * 494; time = 0.0169s; samplesPerSecond = 29239.4
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 2.00598279 * 503; EvalErrorPrediction = 0.52683897 * 503; time = 0.0250s; samplesPerSecond = 20135.3
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96239369 * 487; EvalErrorPrediction = 0.54004107 * 487; time = 0.0167s; samplesPerSecond = 29081.6
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.96513086 * 516; EvalErrorPrediction = 0.51744186 * 516; time = 0.0171s; samplesPerSecond = 30145.5
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 1.97088895 * 494; EvalErrorPrediction = 0.55668016 * 494; time = 0.0169s; samplesPerSecond = 29220.4
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 2.01240053 * 510; EvalErrorPrediction = 0.56470588 * 510; time = 0.0170s; samplesPerSecond = 29947.2
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 2.02387287 * 497; EvalErrorPrediction = 0.54929577 * 497; time = 0.0169s; samplesPerSecond = 29464.1
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.90663199 * 490; EvalErrorPrediction = 0.55306122 * 490; time = 0.0168s; samplesPerSecond = 29196.2
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.94815649 * 482; EvalErrorPrediction = 0.52697095 * 482; time = 0.0167s; samplesPerSecond = 28841.6
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.97332721 * 342; EvalErrorPrediction = 0.50877193 * 342; time = 0.0114s; samplesPerSecond = 29947.5
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 1.93293190 * 497; EvalClassificationError = 0.50905433 * 497; time = 0.0989s; samplesPerSecond = 5024.7
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.13718669 * 492; EvalClassificationError = 0.59552846 * 492; time = 0.0262s; samplesPerSecond = 18811.7
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 1.91004653 * 508; EvalClassificationError = 0.54527559 * 508; time = 0.0266s; samplesPerSecond = 19105.6
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.97341931 * 503; EvalClassificationError = 0.52286282 * 503; time = 0.0264s; samplesPerSecond = 19054.5
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.07837958 * 470; EvalClassificationError = 0.57021277 * 470; time = 0.0165s; samplesPerSecond = 28431.4
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.91466321 * 494; EvalClassificationError = 0.54858300 * 494; time = 0.0169s; samplesPerSecond = 29239.4
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 2.00598279 * 503; EvalClassificationError = 0.52683897 * 503; time = 0.0250s; samplesPerSecond = 20135.3
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96239369 * 487; EvalClassificationError = 0.54004107 * 487; time = 0.0167s; samplesPerSecond = 29081.6
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.96513086 * 516; EvalClassificationError = 0.51744186 * 516; time = 0.0171s; samplesPerSecond = 30145.5
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 1.97088895 * 494; EvalClassificationError = 0.55668016 * 494; time = 0.0169s; samplesPerSecond = 29220.4
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 2.01240053 * 510; EvalClassificationError = 0.56470588 * 510; time = 0.0170s; samplesPerSecond = 29947.2
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 2.02387287 * 497; EvalClassificationError = 0.54929577 * 497; time = 0.0169s; samplesPerSecond = 29464.1
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.90663199 * 490; EvalClassificationError = 0.55306122 * 490; time = 0.0168s; samplesPerSecond = 29196.2
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.94815649 * 482; EvalClassificationError = 0.52697095 * 482; time = 0.0167s; samplesPerSecond = 28841.6
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.97332721 * 342; EvalClassificationError = 0.50877193 * 342; time = 0.0114s; samplesPerSecond = 29947.5
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.15 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.30 seconds since last report (0.01 seconds on comm.); 7692 samples processed by 2 workers (6788 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 25.33k samplesPerSecond , throughputPerWorker = 12.67k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.05172118 * 20480; EvalErrorPrediction = 0.55805664 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.911963s
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.05172118 * 20480; EvalClassificationError = 0.55805664 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.911963s
|
||||
MPI Rank 0: 08/16/2016 03:20:12: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBM@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -718,26 +718,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.96185095 * 1942; EvalErrorPrediction = 0.53347065 * 1942; time = 0.0815s; samplesPerSecond = 23818.3
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.96185095 * 1942; EvalClassificationError = 0.53347065 * 1942; time = 0.0815s; samplesPerSecond = 23818.3
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.07-seconds latency this time; accumulated time on sync point = 0.07 seconds , average latency = 0.07 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.21 seconds since last report (0.01 seconds on comm.); 4885 samples processed by 2 workers (2592 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 23.55k samplesPerSecond , throughputPerWorker = 11.77k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94171416 * 1909; EvalErrorPrediction = 0.55055003 * 1909; time = 0.1664s; samplesPerSecond = 11474.4
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94171416 * 1909; EvalClassificationError = 0.55055003 * 1909; time = 0.1664s; samplesPerSecond = 11474.4
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.07 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.15 seconds since last report (0.01 seconds on comm.); 4826 samples processed by 2 workers (2577 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 33.06k samplesPerSecond , throughputPerWorker = 16.53k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.98907844 * 1987; EvalErrorPrediction = 0.55158530 * 1987; time = 0.1247s; samplesPerSecond = 15928.5
|
||||
MPI Rank 0: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.98907844 * 1987; EvalClassificationError = 0.55158530 * 1987; time = 0.1247s; samplesPerSecond = 15928.5
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.10 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.15 seconds since last report (0.01 seconds on comm.); 4903 samples processed by 2 workers (2577 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 32.81k samplesPerSecond , throughputPerWorker = 16.40k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.94333646 * 1908; EvalErrorPrediction = 0.54926625 * 1908; time = 0.1201s; samplesPerSecond = 15880.5
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.97418902 * 1905; EvalErrorPrediction = 0.55223097 * 1905; time = 0.0580s; samplesPerSecond = 32831.2
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.96248532 * 1913; EvalErrorPrediction = 0.54887611 * 1913; time = 0.0561s; samplesPerSecond = 34070.1
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.97409307 * 1225; EvalErrorPrediction = 0.54367347 * 1225; time = 0.0289s; samplesPerSecond = 42355.3
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.94333646 * 1908; EvalClassificationError = 0.54926625 * 1908; time = 0.1201s; samplesPerSecond = 15880.5
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.97418902 * 1905; EvalClassificationError = 0.55223097 * 1905; time = 0.0580s; samplesPerSecond = 32831.2
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.96248532 * 1913; EvalClassificationError = 0.54887611 * 1913; time = 0.0561s; samplesPerSecond = 34070.1
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.97409307 * 1225; EvalClassificationError = 0.54367347 * 1225; time = 0.0289s; samplesPerSecond = 42355.3
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.10 seconds , average latency = 0.02 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.17 seconds since last report (0.01 seconds on comm.); 5866 samples processed by 2 workers (5043 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 34.19k samplesPerSecond , throughputPerWorker = 17.10k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.95703393 * 20480; EvalErrorPrediction = 0.54541016 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.676426s
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.95703393 * 20480; EvalClassificationError = 0.54541016 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.676426s
|
||||
MPI Rank 0: 08/16/2016 03:20:13: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBM@release_gpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -745,26 +745,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.88963761 * 1923; EvalErrorPrediction = 0.52366095 * 1923; time = 0.0686s; samplesPerSecond = 28024.7
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.88963761 * 1923; EvalClassificationError = 0.52366095 * 1923; time = 0.0686s; samplesPerSecond = 28024.7
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.05-seconds latency this time; accumulated time on sync point = 0.05 seconds , average latency = 0.05 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.17 seconds since last report (0.01 seconds on comm.); 4901 samples processed by 2 workers (2550 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 29.57k samplesPerSecond , throughputPerWorker = 14.78k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.89669303 * 1870; EvalErrorPrediction = 0.51871658 * 1870; time = 0.1373s; samplesPerSecond = 13617.7
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.89669303 * 1870; EvalClassificationError = 0.51871658 * 1870; time = 0.1373s; samplesPerSecond = 13617.7
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.08 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.13 seconds since last report (0.00 seconds on comm.); 4836 samples processed by 2 workers (2519 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 38.34k samplesPerSecond , throughputPerWorker = 19.17k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.91111689 * 1942; EvalErrorPrediction = 0.54119464 * 1942; time = 0.1103s; samplesPerSecond = 17603.8
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.91111689 * 1942; EvalClassificationError = 0.54119464 * 1942; time = 0.1103s; samplesPerSecond = 17603.8
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.11 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.15 seconds since last report (0.00 seconds on comm.); 4952 samples processed by 2 workers (2551 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 33.33k samplesPerSecond , throughputPerWorker = 16.67k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.88475158 * 1885; EvalErrorPrediction = 0.51458886 * 1885; time = 0.1182s; samplesPerSecond = 15950.5
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.89480846 * 1870; EvalErrorPrediction = 0.51497326 * 1870; time = 0.0611s; samplesPerSecond = 30621.6
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.89539137 * 1873; EvalErrorPrediction = 0.52108916 * 1873; time = 0.0441s; samplesPerSecond = 42465.9
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.93128839 * 1231; EvalErrorPrediction = 0.52721365 * 1231; time = 0.0292s; samplesPerSecond = 42189.3
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.88475158 * 1885; EvalClassificationError = 0.51458886 * 1885; time = 0.1182s; samplesPerSecond = 15950.5
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.89480846 * 1870; EvalClassificationError = 0.51497326 * 1870; time = 0.0611s; samplesPerSecond = 30621.6
|
||||
MPI Rank 0: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.89539137 * 1873; EvalClassificationError = 0.52108916 * 1873; time = 0.0441s; samplesPerSecond = 42465.9
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.93128839 * 1231; EvalClassificationError = 0.52721365 * 1231; time = 0.0292s; samplesPerSecond = 42189.3
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.11 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.16 seconds since last report (0.01 seconds on comm.); 5791 samples processed by 2 workers (4974 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 35.98k samplesPerSecond , throughputPerWorker = 17.99k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.90306770 * 20480; EvalErrorPrediction = 0.52641602 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=0.604274s
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.90306770 * 20480; EvalClassificationError = 0.52641602 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=0.604274s
|
||||
MPI Rank 0: 08/16/2016 03:20:14: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBM@release_gpu/models/cntkSpeech.dnn.4'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Starting Epoch 5: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -772,26 +772,26 @@ MPI Rank 0: Parallel training (2 workers) using BlockMomentumSGD with block mome
|
|||
MPI Rank 0: minibatchiterator: epoch 4: frames [81920..102400] (first utterance at frame 81920), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.93021270 * 1863; EvalErrorPrediction = 0.51851852 * 1863; time = 0.0632s; samplesPerSecond = 29459.7
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.93021270 * 1863; EvalClassificationError = 0.51851852 * 1863; time = 0.0632s; samplesPerSecond = 29459.7
|
||||
MPI Rank 0: (model aggregation stats): 1-th sync point was hit, introducing a 0.06-seconds latency this time; accumulated time on sync point = 0.06 seconds , average latency = 0.06 seconds
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: 0.17 seconds since last report (0.01 seconds on comm.); 4919 samples processed by 2 workers (2493 by me);
|
||||
MPI Rank 0: (model aggregation stats) 1-th sync: totalThroughput = 29.52k samplesPerSecond , throughputPerWorker = 14.76k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.87273976 * 1855; EvalErrorPrediction = 0.52129380 * 1855; time = 0.1587s; samplesPerSecond = 11685.4
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.87273976 * 1855; EvalClassificationError = 0.52129380 * 1855; time = 0.1587s; samplesPerSecond = 11685.4
|
||||
MPI Rank 0: (model aggregation stats): 2-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.09 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: 0.14 seconds since last report (0.00 seconds on comm.); 4899 samples processed by 2 workers (2480 by me);
|
||||
MPI Rank 0: (model aggregation stats) 2-th sync: totalThroughput = 35.61k samplesPerSecond , throughputPerWorker = 17.81k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.87195439 * 1866; EvalErrorPrediction = 0.52304394 * 1866; time = 0.1007s; samplesPerSecond = 18521.6
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.87195439 * 1866; EvalClassificationError = 0.52304394 * 1866; time = 0.1007s; samplesPerSecond = 18521.6
|
||||
MPI Rank 0: (model aggregation stats): 3-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.11 seconds , average latency = 0.04 seconds
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: 0.13 seconds since last report (0.00 seconds on comm.); 4829 samples processed by 2 workers (2470 by me);
|
||||
MPI Rank 0: (model aggregation stats) 3-th sync: totalThroughput = 37.08k samplesPerSecond , throughputPerWorker = 18.54k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.88262131 * 1859; EvalErrorPrediction = 0.52017214 * 1859; time = 0.1062s; samplesPerSecond = 17508.2
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.82076948 * 1925; EvalErrorPrediction = 0.49714286 * 1925; time = 0.0664s; samplesPerSecond = 28998.8
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.84625728 * 1860; EvalErrorPrediction = 0.50967742 * 1860; time = 0.0447s; samplesPerSecond = 41652.7
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.86158884 * 1239; EvalErrorPrediction = 0.51412429 * 1239; time = 0.0292s; samplesPerSecond = 42419.9
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.88262131 * 1859; EvalClassificationError = 0.52017214 * 1859; time = 0.1062s; samplesPerSecond = 17508.2
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.82076948 * 1925; EvalClassificationError = 0.49714286 * 1925; time = 0.0664s; samplesPerSecond = 28998.8
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.84625728 * 1860; EvalClassificationError = 0.50967742 * 1860; time = 0.0447s; samplesPerSecond = 41652.7
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.86158884 * 1239; EvalClassificationError = 0.51412429 * 1239; time = 0.0292s; samplesPerSecond = 42419.9
|
||||
MPI Rank 0: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.11 seconds , average latency = 0.03 seconds
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: 0.17 seconds since last report (0.01 seconds on comm.); 5833 samples processed by 2 workers (5024 by me);
|
||||
MPI Rank 0: (model aggregation stats) 4-th sync: totalThroughput = 35.03k samplesPerSecond , throughputPerWorker = 17.51k samplesPerSecond
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.88963745 * 20480; EvalErrorPrediction = 0.51865234 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-005; epochTime=0.603172s
|
||||
MPI Rank 0: 08/16/2016 03:20:14: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.88963745 * 20480; EvalClassificationError = 0.51865234 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-005; epochTime=0.603172s
|
||||
MPI Rank 0: 08/16/2016 03:20:14: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBM@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:20:14: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -845,7 +845,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -871,7 +871,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -953,7 +953,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -979,7 +979,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1067,7 +1067,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1093,7 +1093,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1193,7 +1193,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1222,7 +1222,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1245,7 +1245,7 @@ MPI Rank 1: 08/16/2016 03:20:03: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:20:03: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:03: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:20:03: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:20:03: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1298,237 +1298,237 @@ MPI Rank 1: 08/16/2016 03:20:09: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.68332137 * 192; EvalErrorPrediction = 0.98958333 * 192; time = 0.0394s; samplesPerSecond = 4877.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.42217834 * 192; EvalErrorPrediction = 0.89062500 * 192; time = 0.0213s; samplesPerSecond = 9003.9
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.78361173 * 192; EvalErrorPrediction = 0.93229167 * 192; time = 0.0217s; samplesPerSecond = 8834.9
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.28266373 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0220s; samplesPerSecond = 8730.8
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.48941255 * 192; EvalErrorPrediction = 0.89583333 * 192; time = 0.0218s; samplesPerSecond = 8809.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.63%]: CrossEntropyWithSoftmax = 4.39312485 * 192; EvalErrorPrediction = 0.94270833 * 192; time = 0.0217s; samplesPerSecond = 8858.5
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.26351131 * 192; EvalErrorPrediction = 0.98437500 * 192; time = 0.0214s; samplesPerSecond = 8985.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06459300 * 192; EvalErrorPrediction = 0.90625000 * 192; time = 0.0217s; samplesPerSecond = 8860.6
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97918086 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0208s; samplesPerSecond = 9241.0
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83987406 * 192; EvalErrorPrediction = 0.85416667 * 192; time = 0.0206s; samplesPerSecond = 9337.2
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79421365 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0205s; samplesPerSecond = 9367.7
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.76043875 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0205s; samplesPerSecond = 9377.7
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73154819 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0205s; samplesPerSecond = 9370.9
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.13%]: CrossEntropyWithSoftmax = 3.68435113 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0205s; samplesPerSecond = 9360.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89304289 * 192; EvalErrorPrediction = 0.90104167 * 192; time = 0.0205s; samplesPerSecond = 9371.8
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.76826980 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0206s; samplesPerSecond = 9301.0
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81256010 * 192; EvalErrorPrediction = 0.91145833 * 192; time = 0.0205s; samplesPerSecond = 9371.3
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.92133036 * 192; EvalErrorPrediction = 0.87500000 * 192; time = 0.0205s; samplesPerSecond = 9381.0
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56128851 * 192; EvalErrorPrediction = 0.84895833 * 192; time = 0.0205s; samplesPerSecond = 9369.1
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.67830123 * 192; EvalErrorPrediction = 0.86979167 * 192; time = 0.0203s; samplesPerSecond = 9436.7
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.36612487 * 192; EvalErrorPrediction = 0.76041667 * 192; time = 0.0205s; samplesPerSecond = 9370.9
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.63%]: CrossEntropyWithSoftmax = 3.41785793 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9363.1
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.37660990 * 192; EvalErrorPrediction = 0.79166667 * 192; time = 0.0205s; samplesPerSecond = 9384.6
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.68727319 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0203s; samplesPerSecond = 9439.1
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.64994161 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0205s; samplesPerSecond = 9354.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.37700933 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0204s; samplesPerSecond = 9434.9
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51711918 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0204s; samplesPerSecond = 9397.0
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47828208 * 192; EvalErrorPrediction = 0.80208333 * 192; time = 0.0205s; samplesPerSecond = 9369.1
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.54864536 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0205s; samplesPerSecond = 9369.5
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.13%]: CrossEntropyWithSoftmax = 3.43454656 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9369.1
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.36875092 * 192; EvalErrorPrediction = 0.79687500 * 192; time = 0.0204s; samplesPerSecond = 9393.8
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55401747 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0205s; samplesPerSecond = 9376.8
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.34809828 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9345.3
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.49450178 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9382.3
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.55445642 * 192; EvalErrorPrediction = 0.83333333 * 192; time = 0.0205s; samplesPerSecond = 9360.8
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.49293717 * 192; EvalErrorPrediction = 0.81770833 * 192; time = 0.0204s; samplesPerSecond = 9415.5
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.32494958 * 192; EvalErrorPrediction = 0.79687500 * 192; time = 0.0208s; samplesPerSecond = 9222.3
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.63%]: CrossEntropyWithSoftmax = 3.28851645 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0206s; samplesPerSecond = 9320.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.19411841 * 192; EvalErrorPrediction = 0.78125000 * 192; time = 0.0205s; samplesPerSecond = 9362.7
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.25028950 * 192; EvalErrorPrediction = 0.78125000 * 192; time = 0.0205s; samplesPerSecond = 9379.6
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.53445091 * 192; EvalErrorPrediction = 0.82812500 * 192; time = 0.0205s; samplesPerSecond = 9370.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00326214 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0205s; samplesPerSecond = 9374.1
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.10787619 * 192; EvalErrorPrediction = 0.80729167 * 192; time = 0.0205s; samplesPerSecond = 9359.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.05588285 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0205s; samplesPerSecond = 9361.7
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18197888 * 192; EvalErrorPrediction = 0.74479167 * 192; time = 0.0205s; samplesPerSecond = 9359.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.13%]: CrossEntropyWithSoftmax = 2.78138232 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0204s; samplesPerSecond = 9394.3
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17441293 * 192; EvalErrorPrediction = 0.82291667 * 192; time = 0.0205s; samplesPerSecond = 9364.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03537426 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0205s; samplesPerSecond = 9363.1
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.00595982 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0205s; samplesPerSecond = 9375.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79115323 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9370.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.04097805 * 192; EvalErrorPrediction = 0.69791667 * 192; time = 0.0205s; samplesPerSecond = 9373.2
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.15935323 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0205s; samplesPerSecond = 9373.6
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11947251 * 192; EvalErrorPrediction = 0.75000000 * 192; time = 0.0205s; samplesPerSecond = 9363.6
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.63%]: CrossEntropyWithSoftmax = 2.86232384 * 192; EvalErrorPrediction = 0.72395833 * 192; time = 0.0205s; samplesPerSecond = 9364.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.77742114 * 192; EvalErrorPrediction = 0.70312500 * 192; time = 0.0205s; samplesPerSecond = 9376.8
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00411374 * 192; EvalErrorPrediction = 0.73958333 * 192; time = 0.0204s; samplesPerSecond = 9409.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69740857 * 192; EvalErrorPrediction = 0.68750000 * 192; time = 0.0205s; samplesPerSecond = 9371.3
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64948203 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0204s; samplesPerSecond = 9399.8
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71417618 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0205s; samplesPerSecond = 9355.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.74281938 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0204s; samplesPerSecond = 9400.2
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.81346277 * 192; EvalErrorPrediction = 0.72916667 * 192; time = 0.0206s; samplesPerSecond = 9339.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79862588 * 192; EvalErrorPrediction = 0.71875000 * 192; time = 0.0205s; samplesPerSecond = 9364.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.76655026 * 192; EvalErrorPrediction = 0.71354167 * 192; time = 0.0205s; samplesPerSecond = 9370.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.75908943 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0204s; samplesPerSecond = 9420.1
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.53548199 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0205s; samplesPerSecond = 9365.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97589950 * 192; EvalErrorPrediction = 0.73437500 * 192; time = 0.0205s; samplesPerSecond = 9355.8
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.64996722 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0205s; samplesPerSecond = 9375.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52128374 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9377.7
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.64228785 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0205s; samplesPerSecond = 9358.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.63%]: CrossEntropyWithSoftmax = 2.57199182 * 192; EvalErrorPrediction = 0.67708333 * 192; time = 0.0205s; samplesPerSecond = 9364.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.58100902 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0205s; samplesPerSecond = 9370.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.48555431 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0205s; samplesPerSecond = 9358.1
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.75336137 * 192; EvalErrorPrediction = 0.67187500 * 192; time = 0.0205s; samplesPerSecond = 9375.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.49193178 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0205s; samplesPerSecond = 9373.2
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.46098647 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0205s; samplesPerSecond = 9372.3
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.74322101 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0205s; samplesPerSecond = 9361.3
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.55837089 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0205s; samplesPerSecond = 9360.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.13%]: CrossEntropyWithSoftmax = 2.98288510 * 192; EvalErrorPrediction = 0.76562500 * 192; time = 0.0204s; samplesPerSecond = 9406.7
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.36667287 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0205s; samplesPerSecond = 9376.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.25169614 * 192; EvalErrorPrediction = 0.61458333 * 192; time = 0.0205s; samplesPerSecond = 9352.2
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.31564120 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0205s; samplesPerSecond = 9366.3
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.70894816 * 192; EvalErrorPrediction = 0.70833333 * 192; time = 0.0206s; samplesPerSecond = 9341.7
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.44991146 * 192; EvalErrorPrediction = 0.63020833 * 192; time = 0.0205s; samplesPerSecond = 9369.1
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.51856232 * 192; EvalErrorPrediction = 0.66666667 * 192; time = 0.0205s; samplesPerSecond = 9375.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.38498228 * 192; EvalErrorPrediction = 0.61979167 * 192; time = 0.0205s; samplesPerSecond = 9373.6
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.63%]: CrossEntropyWithSoftmax = 2.38080698 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0205s; samplesPerSecond = 9364.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.34294967 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9355.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.30340167 * 192; EvalErrorPrediction = 0.58854167 * 192; time = 0.0204s; samplesPerSecond = 9391.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.08323277 * 192; EvalErrorPrediction = 0.52604167 * 192; time = 0.0206s; samplesPerSecond = 9342.2
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.43589953 * 192; EvalErrorPrediction = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9368.1
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.53399486 * 192; EvalErrorPrediction = 0.66145833 * 192; time = 0.0205s; samplesPerSecond = 9364.5
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33995708 * 192; EvalErrorPrediction = 0.60416667 * 192; time = 0.0205s; samplesPerSecond = 9364.0
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.62970864 * 192; EvalErrorPrediction = 0.68229167 * 192; time = 0.0205s; samplesPerSecond = 9360.4
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.13%]: CrossEntropyWithSoftmax = 2.47993989 * 192; EvalErrorPrediction = 0.64062500 * 192; time = 0.0205s; samplesPerSecond = 9368.6
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30935254 * 192; EvalErrorPrediction = 0.63541667 * 192; time = 0.0205s; samplesPerSecond = 9378.7
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.22022265 * 192; EvalErrorPrediction = 0.59375000 * 192; time = 0.0205s; samplesPerSecond = 9379.1
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.28060247 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0205s; samplesPerSecond = 9367.7
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.13349744 * 192; EvalErrorPrediction = 0.57291667 * 192; time = 0.0204s; samplesPerSecond = 9417.3
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.29751600 * 192; EvalErrorPrediction = 0.65104167 * 192; time = 0.0205s; samplesPerSecond = 9370.9
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.29319143 * 192; EvalErrorPrediction = 0.60416667 * 192; time = 0.0204s; samplesPerSecond = 9414.5
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.14551002 * 192; EvalErrorPrediction = 0.55729167 * 192; time = 0.0205s; samplesPerSecond = 9377.7
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.63%]: CrossEntropyWithSoftmax = 2.26930674 * 192; EvalErrorPrediction = 0.58333333 * 192; time = 0.0204s; samplesPerSecond = 9397.9
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.17383355 * 192; EvalErrorPrediction = 0.56770833 * 192; time = 0.0205s; samplesPerSecond = 9380.0
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.43111882 * 192; EvalErrorPrediction = 0.67187500 * 192; time = 0.0205s; samplesPerSecond = 9373.6
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.18011227 * 192; EvalErrorPrediction = 0.59895833 * 192; time = 0.0205s; samplesPerSecond = 9378.2
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.21682707 * 192; EvalErrorPrediction = 0.56250000 * 192; time = 0.0174s; samplesPerSecond = 11050.4
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.25919s
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 1- 3, 0.94%]: CrossEntropyWithSoftmax = 4.68332137 * 192; EvalClassificationError = 0.98958333 * 192; time = 0.0394s; samplesPerSecond = 4877.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 4- 6, 1.88%]: CrossEntropyWithSoftmax = 4.42217834 * 192; EvalClassificationError = 0.89062500 * 192; time = 0.0213s; samplesPerSecond = 9003.9
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 7- 9, 2.81%]: CrossEntropyWithSoftmax = 4.78361173 * 192; EvalClassificationError = 0.93229167 * 192; time = 0.0217s; samplesPerSecond = 8834.9
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 10- 12, 3.75%]: CrossEntropyWithSoftmax = 4.28266373 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0220s; samplesPerSecond = 8730.8
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 13- 15, 4.69%]: CrossEntropyWithSoftmax = 4.48941255 * 192; EvalClassificationError = 0.89583333 * 192; time = 0.0218s; samplesPerSecond = 8809.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 16- 18, 5.63%]: CrossEntropyWithSoftmax = 4.39312485 * 192; EvalClassificationError = 0.94270833 * 192; time = 0.0217s; samplesPerSecond = 8858.5
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 19- 21, 6.56%]: CrossEntropyWithSoftmax = 4.26351131 * 192; EvalClassificationError = 0.98437500 * 192; time = 0.0214s; samplesPerSecond = 8985.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 22- 24, 7.50%]: CrossEntropyWithSoftmax = 4.06459300 * 192; EvalClassificationError = 0.90625000 * 192; time = 0.0217s; samplesPerSecond = 8860.6
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 25- 27, 8.44%]: CrossEntropyWithSoftmax = 3.97918086 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0208s; samplesPerSecond = 9241.0
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 28- 30, 9.38%]: CrossEntropyWithSoftmax = 3.83987406 * 192; EvalClassificationError = 0.85416667 * 192; time = 0.0206s; samplesPerSecond = 9337.2
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 31- 33, 10.31%]: CrossEntropyWithSoftmax = 3.79421365 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0205s; samplesPerSecond = 9367.7
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 34- 36, 11.25%]: CrossEntropyWithSoftmax = 3.76043875 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0205s; samplesPerSecond = 9377.7
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 37- 39, 12.19%]: CrossEntropyWithSoftmax = 3.73154819 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0205s; samplesPerSecond = 9370.9
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 40- 42, 13.13%]: CrossEntropyWithSoftmax = 3.68435113 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0205s; samplesPerSecond = 9360.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 43- 45, 14.06%]: CrossEntropyWithSoftmax = 3.89304289 * 192; EvalClassificationError = 0.90104167 * 192; time = 0.0205s; samplesPerSecond = 9371.8
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 46- 48, 15.00%]: CrossEntropyWithSoftmax = 3.76826980 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0206s; samplesPerSecond = 9301.0
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 49- 51, 15.94%]: CrossEntropyWithSoftmax = 3.81256010 * 192; EvalClassificationError = 0.91145833 * 192; time = 0.0205s; samplesPerSecond = 9371.3
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 52- 54, 16.88%]: CrossEntropyWithSoftmax = 3.92133036 * 192; EvalClassificationError = 0.87500000 * 192; time = 0.0205s; samplesPerSecond = 9381.0
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 55- 57, 17.81%]: CrossEntropyWithSoftmax = 3.56128851 * 192; EvalClassificationError = 0.84895833 * 192; time = 0.0205s; samplesPerSecond = 9369.1
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 58- 60, 18.75%]: CrossEntropyWithSoftmax = 3.67830123 * 192; EvalClassificationError = 0.86979167 * 192; time = 0.0203s; samplesPerSecond = 9436.7
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 61- 63, 19.69%]: CrossEntropyWithSoftmax = 3.36612487 * 192; EvalClassificationError = 0.76041667 * 192; time = 0.0205s; samplesPerSecond = 9370.9
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 64- 66, 20.63%]: CrossEntropyWithSoftmax = 3.41785793 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9363.1
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 67- 69, 21.56%]: CrossEntropyWithSoftmax = 3.37660990 * 192; EvalClassificationError = 0.79166667 * 192; time = 0.0205s; samplesPerSecond = 9384.6
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 70- 72, 22.50%]: CrossEntropyWithSoftmax = 3.68727319 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0203s; samplesPerSecond = 9439.1
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 73- 75, 23.44%]: CrossEntropyWithSoftmax = 3.64994161 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0205s; samplesPerSecond = 9354.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 76- 78, 24.38%]: CrossEntropyWithSoftmax = 3.37700933 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0204s; samplesPerSecond = 9434.9
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 79- 81, 25.31%]: CrossEntropyWithSoftmax = 3.51711918 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0204s; samplesPerSecond = 9397.0
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 82- 84, 26.25%]: CrossEntropyWithSoftmax = 3.47828208 * 192; EvalClassificationError = 0.80208333 * 192; time = 0.0205s; samplesPerSecond = 9369.1
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 85- 87, 27.19%]: CrossEntropyWithSoftmax = 3.54864536 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0205s; samplesPerSecond = 9369.5
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 88- 90, 28.13%]: CrossEntropyWithSoftmax = 3.43454656 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9369.1
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 91- 93, 29.06%]: CrossEntropyWithSoftmax = 3.36875092 * 192; EvalClassificationError = 0.79687500 * 192; time = 0.0204s; samplesPerSecond = 9393.8
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 94- 96, 30.00%]: CrossEntropyWithSoftmax = 3.55401747 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0205s; samplesPerSecond = 9376.8
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 97- 99, 30.94%]: CrossEntropyWithSoftmax = 3.34809828 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9345.3
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 100- 102, 31.87%]: CrossEntropyWithSoftmax = 3.49450178 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0205s; samplesPerSecond = 9382.3
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 103- 105, 32.81%]: CrossEntropyWithSoftmax = 3.55445642 * 192; EvalClassificationError = 0.83333333 * 192; time = 0.0205s; samplesPerSecond = 9360.8
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 106- 108, 33.75%]: CrossEntropyWithSoftmax = 3.49293717 * 192; EvalClassificationError = 0.81770833 * 192; time = 0.0204s; samplesPerSecond = 9415.5
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 109- 111, 34.69%]: CrossEntropyWithSoftmax = 3.32494958 * 192; EvalClassificationError = 0.79687500 * 192; time = 0.0208s; samplesPerSecond = 9222.3
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 112- 114, 35.63%]: CrossEntropyWithSoftmax = 3.28851645 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0206s; samplesPerSecond = 9320.4
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 115- 117, 36.56%]: CrossEntropyWithSoftmax = 3.19411841 * 192; EvalClassificationError = 0.78125000 * 192; time = 0.0205s; samplesPerSecond = 9362.7
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 118- 120, 37.50%]: CrossEntropyWithSoftmax = 3.25028950 * 192; EvalClassificationError = 0.78125000 * 192; time = 0.0205s; samplesPerSecond = 9379.6
|
||||
MPI Rank 1: 08/16/2016 03:20:09: Epoch[ 1 of 5]-Minibatch[ 121- 123, 38.44%]: CrossEntropyWithSoftmax = 3.53445091 * 192; EvalClassificationError = 0.82812500 * 192; time = 0.0205s; samplesPerSecond = 9370.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 124- 126, 39.38%]: CrossEntropyWithSoftmax = 3.00326214 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0205s; samplesPerSecond = 9374.1
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 127- 129, 40.31%]: CrossEntropyWithSoftmax = 3.10787619 * 192; EvalClassificationError = 0.80729167 * 192; time = 0.0205s; samplesPerSecond = 9359.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 130- 132, 41.25%]: CrossEntropyWithSoftmax = 3.05588285 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0205s; samplesPerSecond = 9361.7
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 133- 135, 42.19%]: CrossEntropyWithSoftmax = 3.18197888 * 192; EvalClassificationError = 0.74479167 * 192; time = 0.0205s; samplesPerSecond = 9359.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 136- 138, 43.13%]: CrossEntropyWithSoftmax = 2.78138232 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0204s; samplesPerSecond = 9394.3
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 139- 141, 44.06%]: CrossEntropyWithSoftmax = 3.17441293 * 192; EvalClassificationError = 0.82291667 * 192; time = 0.0205s; samplesPerSecond = 9364.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 142- 144, 45.00%]: CrossEntropyWithSoftmax = 3.03537426 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0205s; samplesPerSecond = 9363.1
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 145- 147, 45.94%]: CrossEntropyWithSoftmax = 3.00595982 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0205s; samplesPerSecond = 9375.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 148- 150, 46.88%]: CrossEntropyWithSoftmax = 2.79115323 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9370.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 151- 153, 47.81%]: CrossEntropyWithSoftmax = 3.04097805 * 192; EvalClassificationError = 0.69791667 * 192; time = 0.0205s; samplesPerSecond = 9373.2
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 154- 156, 48.75%]: CrossEntropyWithSoftmax = 3.15935323 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0205s; samplesPerSecond = 9373.6
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 157- 159, 49.69%]: CrossEntropyWithSoftmax = 3.11947251 * 192; EvalClassificationError = 0.75000000 * 192; time = 0.0205s; samplesPerSecond = 9363.6
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 160- 162, 50.63%]: CrossEntropyWithSoftmax = 2.86232384 * 192; EvalClassificationError = 0.72395833 * 192; time = 0.0205s; samplesPerSecond = 9364.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 163- 165, 51.56%]: CrossEntropyWithSoftmax = 2.77742114 * 192; EvalClassificationError = 0.70312500 * 192; time = 0.0205s; samplesPerSecond = 9376.8
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 166- 168, 52.50%]: CrossEntropyWithSoftmax = 3.00411374 * 192; EvalClassificationError = 0.73958333 * 192; time = 0.0204s; samplesPerSecond = 9409.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 169- 171, 53.44%]: CrossEntropyWithSoftmax = 2.69740857 * 192; EvalClassificationError = 0.68750000 * 192; time = 0.0205s; samplesPerSecond = 9371.3
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 172- 174, 54.37%]: CrossEntropyWithSoftmax = 2.64948203 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0204s; samplesPerSecond = 9399.8
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 175- 177, 55.31%]: CrossEntropyWithSoftmax = 2.71417618 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0205s; samplesPerSecond = 9355.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 178- 180, 56.25%]: CrossEntropyWithSoftmax = 2.74281938 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0204s; samplesPerSecond = 9400.2
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 181- 183, 57.19%]: CrossEntropyWithSoftmax = 2.81346277 * 192; EvalClassificationError = 0.72916667 * 192; time = 0.0206s; samplesPerSecond = 9339.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 184- 186, 58.13%]: CrossEntropyWithSoftmax = 2.79862588 * 192; EvalClassificationError = 0.71875000 * 192; time = 0.0205s; samplesPerSecond = 9364.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 187- 189, 59.06%]: CrossEntropyWithSoftmax = 2.76655026 * 192; EvalClassificationError = 0.71354167 * 192; time = 0.0205s; samplesPerSecond = 9370.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 190- 192, 60.00%]: CrossEntropyWithSoftmax = 2.75908943 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0204s; samplesPerSecond = 9420.1
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 193- 195, 60.94%]: CrossEntropyWithSoftmax = 2.53548199 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0205s; samplesPerSecond = 9365.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 196- 198, 61.88%]: CrossEntropyWithSoftmax = 2.97589950 * 192; EvalClassificationError = 0.73437500 * 192; time = 0.0205s; samplesPerSecond = 9355.8
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 199- 201, 62.81%]: CrossEntropyWithSoftmax = 2.64996722 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0205s; samplesPerSecond = 9375.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 202- 204, 63.75%]: CrossEntropyWithSoftmax = 2.52128374 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9377.7
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 205- 207, 64.69%]: CrossEntropyWithSoftmax = 2.64228785 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0205s; samplesPerSecond = 9358.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 208- 210, 65.63%]: CrossEntropyWithSoftmax = 2.57199182 * 192; EvalClassificationError = 0.67708333 * 192; time = 0.0205s; samplesPerSecond = 9364.9
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 211- 213, 66.56%]: CrossEntropyWithSoftmax = 2.58100902 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0205s; samplesPerSecond = 9370.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 214- 216, 67.50%]: CrossEntropyWithSoftmax = 2.48555431 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0205s; samplesPerSecond = 9358.1
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 217- 219, 68.44%]: CrossEntropyWithSoftmax = 2.75336137 * 192; EvalClassificationError = 0.67187500 * 192; time = 0.0205s; samplesPerSecond = 9375.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 220- 222, 69.38%]: CrossEntropyWithSoftmax = 2.49193178 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0205s; samplesPerSecond = 9373.2
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 223- 225, 70.31%]: CrossEntropyWithSoftmax = 2.46098647 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0205s; samplesPerSecond = 9372.3
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 226- 228, 71.25%]: CrossEntropyWithSoftmax = 2.74322101 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0205s; samplesPerSecond = 9361.3
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 229- 231, 72.19%]: CrossEntropyWithSoftmax = 2.55837089 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0205s; samplesPerSecond = 9360.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 232- 234, 73.13%]: CrossEntropyWithSoftmax = 2.98288510 * 192; EvalClassificationError = 0.76562500 * 192; time = 0.0204s; samplesPerSecond = 9406.7
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 235- 237, 74.06%]: CrossEntropyWithSoftmax = 2.36667287 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0205s; samplesPerSecond = 9376.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 238- 240, 75.00%]: CrossEntropyWithSoftmax = 2.25169614 * 192; EvalClassificationError = 0.61458333 * 192; time = 0.0205s; samplesPerSecond = 9352.2
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 241- 243, 75.94%]: CrossEntropyWithSoftmax = 2.31564120 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0205s; samplesPerSecond = 9366.3
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 244- 246, 76.88%]: CrossEntropyWithSoftmax = 2.70894816 * 192; EvalClassificationError = 0.70833333 * 192; time = 0.0206s; samplesPerSecond = 9341.7
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 247- 249, 77.81%]: CrossEntropyWithSoftmax = 2.44991146 * 192; EvalClassificationError = 0.63020833 * 192; time = 0.0205s; samplesPerSecond = 9369.1
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 250- 252, 78.75%]: CrossEntropyWithSoftmax = 2.51856232 * 192; EvalClassificationError = 0.66666667 * 192; time = 0.0205s; samplesPerSecond = 9375.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 253- 255, 79.69%]: CrossEntropyWithSoftmax = 2.38498228 * 192; EvalClassificationError = 0.61979167 * 192; time = 0.0205s; samplesPerSecond = 9373.6
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 256- 258, 80.63%]: CrossEntropyWithSoftmax = 2.38080698 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0205s; samplesPerSecond = 9364.5
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 259- 261, 81.56%]: CrossEntropyWithSoftmax = 2.34294967 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9355.4
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 262- 264, 82.50%]: CrossEntropyWithSoftmax = 2.30340167 * 192; EvalClassificationError = 0.58854167 * 192; time = 0.0204s; samplesPerSecond = 9391.0
|
||||
MPI Rank 1: 08/16/2016 03:20:10: Epoch[ 1 of 5]-Minibatch[ 265- 267, 83.44%]: CrossEntropyWithSoftmax = 2.08323277 * 192; EvalClassificationError = 0.52604167 * 192; time = 0.0206s; samplesPerSecond = 9342.2
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 268- 270, 84.38%]: CrossEntropyWithSoftmax = 2.43589953 * 192; EvalClassificationError = 0.64583333 * 192; time = 0.0205s; samplesPerSecond = 9368.1
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 271- 273, 85.31%]: CrossEntropyWithSoftmax = 2.53399486 * 192; EvalClassificationError = 0.66145833 * 192; time = 0.0205s; samplesPerSecond = 9364.5
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 274- 276, 86.25%]: CrossEntropyWithSoftmax = 2.33995708 * 192; EvalClassificationError = 0.60416667 * 192; time = 0.0205s; samplesPerSecond = 9364.0
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 277- 279, 87.19%]: CrossEntropyWithSoftmax = 2.62970864 * 192; EvalClassificationError = 0.68229167 * 192; time = 0.0205s; samplesPerSecond = 9360.4
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 280- 282, 88.13%]: CrossEntropyWithSoftmax = 2.47993989 * 192; EvalClassificationError = 0.64062500 * 192; time = 0.0205s; samplesPerSecond = 9368.6
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 283- 285, 89.06%]: CrossEntropyWithSoftmax = 2.30935254 * 192; EvalClassificationError = 0.63541667 * 192; time = 0.0205s; samplesPerSecond = 9378.7
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 286- 288, 90.00%]: CrossEntropyWithSoftmax = 2.22022265 * 192; EvalClassificationError = 0.59375000 * 192; time = 0.0205s; samplesPerSecond = 9379.1
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 289- 291, 90.94%]: CrossEntropyWithSoftmax = 2.28060247 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0205s; samplesPerSecond = 9367.7
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 292- 294, 91.88%]: CrossEntropyWithSoftmax = 2.13349744 * 192; EvalClassificationError = 0.57291667 * 192; time = 0.0204s; samplesPerSecond = 9417.3
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 295- 297, 92.81%]: CrossEntropyWithSoftmax = 2.29751600 * 192; EvalClassificationError = 0.65104167 * 192; time = 0.0205s; samplesPerSecond = 9370.9
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 298- 300, 93.75%]: CrossEntropyWithSoftmax = 2.29319143 * 192; EvalClassificationError = 0.60416667 * 192; time = 0.0204s; samplesPerSecond = 9414.5
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 301- 303, 94.69%]: CrossEntropyWithSoftmax = 2.14551002 * 192; EvalClassificationError = 0.55729167 * 192; time = 0.0205s; samplesPerSecond = 9377.7
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 304- 306, 95.63%]: CrossEntropyWithSoftmax = 2.26930674 * 192; EvalClassificationError = 0.58333333 * 192; time = 0.0204s; samplesPerSecond = 9397.9
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 307- 309, 96.56%]: CrossEntropyWithSoftmax = 2.17383355 * 192; EvalClassificationError = 0.56770833 * 192; time = 0.0205s; samplesPerSecond = 9380.0
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 310- 312, 97.50%]: CrossEntropyWithSoftmax = 2.43111882 * 192; EvalClassificationError = 0.67187500 * 192; time = 0.0205s; samplesPerSecond = 9373.6
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 313- 315, 98.44%]: CrossEntropyWithSoftmax = 2.18011227 * 192; EvalClassificationError = 0.59895833 * 192; time = 0.0205s; samplesPerSecond = 9378.2
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 1 of 5]-Minibatch[ 316- 318, 99.38%]: CrossEntropyWithSoftmax = 2.21682707 * 192; EvalClassificationError = 0.56250000 * 192; time = 0.0174s; samplesPerSecond = 11050.4
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Finished Epoch[ 1 of 5]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.25919s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.19272896 * 260; EvalErrorPrediction = 0.61538462 * 260; time = 0.0647s; samplesPerSecond = 4017.3
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.34922865 * 276; EvalErrorPrediction = 0.70289855 * 276; time = 0.0218s; samplesPerSecond = 12672.8
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.24157888 * 280; EvalErrorPrediction = 0.67500000 * 280; time = 0.0264s; samplesPerSecond = 10620.9
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.20817680 * 241; EvalErrorPrediction = 0.62655602 * 241; time = 0.0164s; samplesPerSecond = 14707.7
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.04015325 * 295; EvalErrorPrediction = 0.52881356 * 295; time = 0.0175s; samplesPerSecond = 16844.6
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.40322904 * 257; EvalErrorPrediction = 0.63035019 * 257; time = 0.0127s; samplesPerSecond = 20217.1
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.04484358 * 262; EvalErrorPrediction = 0.56488550 * 262; time = 0.0128s; samplesPerSecond = 20502.4
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 1- 3, 3.75%]: CrossEntropyWithSoftmax = 2.19272896 * 260; EvalClassificationError = 0.61538462 * 260; time = 0.0647s; samplesPerSecond = 4017.3
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 4- 6, 7.50%]: CrossEntropyWithSoftmax = 2.34922865 * 276; EvalClassificationError = 0.70289855 * 276; time = 0.0218s; samplesPerSecond = 12672.8
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 7- 9, 11.25%]: CrossEntropyWithSoftmax = 2.24157888 * 280; EvalClassificationError = 0.67500000 * 280; time = 0.0264s; samplesPerSecond = 10620.9
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 10- 12, 15.00%]: CrossEntropyWithSoftmax = 2.20817680 * 241; EvalClassificationError = 0.62655602 * 241; time = 0.0164s; samplesPerSecond = 14707.7
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 13- 15, 18.75%]: CrossEntropyWithSoftmax = 2.04015325 * 295; EvalClassificationError = 0.52881356 * 295; time = 0.0175s; samplesPerSecond = 16844.6
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 16- 18, 22.50%]: CrossEntropyWithSoftmax = 2.40322904 * 257; EvalClassificationError = 0.63035019 * 257; time = 0.0127s; samplesPerSecond = 20217.1
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 19- 21, 26.25%]: CrossEntropyWithSoftmax = 2.04484358 * 262; EvalClassificationError = 0.56488550 * 262; time = 0.0128s; samplesPerSecond = 20502.4
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.24 seconds since last report (0.01 seconds on comm.); 4289 samples processed by 2 workers (2126 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 18.06k samplesPerSecond , throughputPerWorker = 9.03k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.03646309 * 255; EvalErrorPrediction = 0.53333333 * 255; time = 0.0503s; samplesPerSecond = 5068.3
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.07321525 * 279; EvalErrorPrediction = 0.58781362 * 279; time = 0.0222s; samplesPerSecond = 12592.0
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.06339373 * 274; EvalErrorPrediction = 0.52554745 * 274; time = 0.0269s; samplesPerSecond = 10199.9
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 1.95530128 * 269; EvalErrorPrediction = 0.53531599 * 269; time = 0.0267s; samplesPerSecond = 10070.8
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.03493597 * 278; EvalErrorPrediction = 0.61151079 * 278; time = 0.0267s; samplesPerSecond = 10416.3
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 2.13309941 * 271; EvalErrorPrediction = 0.54612546 * 271; time = 0.0214s; samplesPerSecond = 12692.0
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.19775096 * 276; EvalErrorPrediction = 0.58333333 * 276; time = 0.0129s; samplesPerSecond = 21435.2
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 2.14623378 * 260; EvalErrorPrediction = 0.58076923 * 260; time = 0.0127s; samplesPerSecond = 20496.6
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 22- 24, 30.00%]: CrossEntropyWithSoftmax = 2.03646309 * 255; EvalClassificationError = 0.53333333 * 255; time = 0.0503s; samplesPerSecond = 5068.3
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 25- 27, 33.75%]: CrossEntropyWithSoftmax = 2.07321525 * 279; EvalClassificationError = 0.58781362 * 279; time = 0.0222s; samplesPerSecond = 12592.0
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 28- 30, 37.50%]: CrossEntropyWithSoftmax = 2.06339373 * 274; EvalClassificationError = 0.52554745 * 274; time = 0.0269s; samplesPerSecond = 10199.9
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 31- 33, 41.25%]: CrossEntropyWithSoftmax = 1.95530128 * 269; EvalClassificationError = 0.53531599 * 269; time = 0.0267s; samplesPerSecond = 10070.8
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 34- 36, 45.00%]: CrossEntropyWithSoftmax = 2.03493597 * 278; EvalClassificationError = 0.61151079 * 278; time = 0.0267s; samplesPerSecond = 10416.3
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 37- 39, 48.75%]: CrossEntropyWithSoftmax = 2.13309941 * 271; EvalClassificationError = 0.54612546 * 271; time = 0.0214s; samplesPerSecond = 12692.0
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 40- 42, 52.50%]: CrossEntropyWithSoftmax = 2.19775096 * 276; EvalClassificationError = 0.58333333 * 276; time = 0.0129s; samplesPerSecond = 21435.2
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 43- 45, 56.25%]: CrossEntropyWithSoftmax = 2.14623378 * 260; EvalClassificationError = 0.58076923 * 260; time = 0.0127s; samplesPerSecond = 20496.6
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.18 seconds since last report (0.01 seconds on comm.); 4253 samples processed by 2 workers (2073 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 23.55k samplesPerSecond , throughputPerWorker = 11.77k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.98115282 * 265; EvalErrorPrediction = 0.54716981 * 265; time = 0.0368s; samplesPerSecond = 7192.1
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.13033706 * 298; EvalErrorPrediction = 0.57718121 * 298; time = 0.0273s; samplesPerSecond = 10911.8
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.96671704 * 274; EvalErrorPrediction = 0.50364964 * 274; time = 0.0269s; samplesPerSecond = 10169.2
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.98515695 * 265; EvalErrorPrediction = 0.53962264 * 265; time = 0.0267s; samplesPerSecond = 9929.9
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.85824670 * 281; EvalErrorPrediction = 0.54448399 * 281; time = 0.0148s; samplesPerSecond = 18949.4
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.95611759 * 252; EvalErrorPrediction = 0.52777778 * 252; time = 0.0207s; samplesPerSecond = 12163.9
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 2.04102834 * 274; EvalErrorPrediction = 0.51094891 * 274; time = 0.0166s; samplesPerSecond = 16512.0
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.84439304 * 258; EvalErrorPrediction = 0.49224806 * 258; time = 0.0126s; samplesPerSecond = 20458.3
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 46- 48, 60.00%]: CrossEntropyWithSoftmax = 1.98115282 * 265; EvalClassificationError = 0.54716981 * 265; time = 0.0368s; samplesPerSecond = 7192.1
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 49- 51, 63.75%]: CrossEntropyWithSoftmax = 2.13033706 * 298; EvalClassificationError = 0.57718121 * 298; time = 0.0273s; samplesPerSecond = 10911.8
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 52- 54, 67.50%]: CrossEntropyWithSoftmax = 1.96671704 * 274; EvalClassificationError = 0.50364964 * 274; time = 0.0269s; samplesPerSecond = 10169.2
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 55- 57, 71.25%]: CrossEntropyWithSoftmax = 1.98515695 * 265; EvalClassificationError = 0.53962264 * 265; time = 0.0267s; samplesPerSecond = 9929.9
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 58- 60, 75.00%]: CrossEntropyWithSoftmax = 1.85824670 * 281; EvalClassificationError = 0.54448399 * 281; time = 0.0148s; samplesPerSecond = 18949.4
|
||||
MPI Rank 1: 08/16/2016 03:20:11: Epoch[ 2 of 5]-Minibatch[ 61- 63, 78.75%]: CrossEntropyWithSoftmax = 1.95611759 * 252; EvalClassificationError = 0.52777778 * 252; time = 0.0207s; samplesPerSecond = 12163.9
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 64- 66, 82.50%]: CrossEntropyWithSoftmax = 2.04102834 * 274; EvalClassificationError = 0.51094891 * 274; time = 0.0166s; samplesPerSecond = 16512.0
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 67- 69, 86.25%]: CrossEntropyWithSoftmax = 1.84439304 * 258; EvalClassificationError = 0.49224806 * 258; time = 0.0126s; samplesPerSecond = 20458.3
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.19 seconds since last report (0.01 seconds on comm.); 4246 samples processed by 2 workers (2102 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 22.76k samplesPerSecond , throughputPerWorker = 11.38k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 1.97159866 * 271; EvalErrorPrediction = 0.55719557 * 271; time = 0.0365s; samplesPerSecond = 7430.2
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.97632511 * 278; EvalErrorPrediction = 0.52877698 * 278; time = 0.0266s; samplesPerSecond = 10459.0
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.95095600 * 286; EvalErrorPrediction = 0.56293706 * 286; time = 0.0268s; samplesPerSecond = 10683.6
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.94334189 * 170; EvalErrorPrediction = 0.52941176 * 170; time = 0.0221s; samplesPerSecond = 7693.4
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 70- 72, 90.00%]: CrossEntropyWithSoftmax = 1.97159866 * 271; EvalClassificationError = 0.55719557 * 271; time = 0.0365s; samplesPerSecond = 7430.2
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 73- 75, 93.75%]: CrossEntropyWithSoftmax = 1.97632511 * 278; EvalClassificationError = 0.52877698 * 278; time = 0.0266s; samplesPerSecond = 10459.0
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 76- 78, 97.50%]: CrossEntropyWithSoftmax = 1.95095600 * 286; EvalClassificationError = 0.56293706 * 286; time = 0.0268s; samplesPerSecond = 10683.6
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 2 of 5]-Minibatch[ 79- 81, 101.25%]: CrossEntropyWithSoftmax = 1.94334189 * 170; EvalClassificationError = 0.52941176 * 170; time = 0.0221s; samplesPerSecond = 7693.4
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.30 seconds since last report (0.18 seconds on comm.); 7692 samples processed by 2 workers (904 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 25.25k samplesPerSecond , throughputPerWorker = 12.62k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.05172118 * 20480; EvalErrorPrediction = 0.55805664 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.91203s
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Finished Epoch[ 2 of 5]: [Training] CrossEntropyWithSoftmax = 2.05172118 * 20480; EvalClassificationError = 0.55805664 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.91203s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.94144328 * 1130; EvalErrorPrediction = 0.53097345 * 1130; time = 0.1167s; samplesPerSecond = 9679.1
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.94144328 * 1130; EvalClassificationError = 0.53097345 * 1130; time = 0.1167s; samplesPerSecond = 9679.1
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.21 seconds since last report (0.03 seconds on comm.); 4885 samples processed by 2 workers (2293 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 23.56k samplesPerSecond , throughputPerWorker = 11.78k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.87694957 * 1163; EvalErrorPrediction = 0.52794497 * 1163; time = 0.0816s; samplesPerSecond = 14256.6
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.98228580 * 1085; EvalErrorPrediction = 0.52718894 * 1085; time = 0.0712s; samplesPerSecond = 15246.5
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.87694957 * 1163; EvalClassificationError = 0.52794497 * 1163; time = 0.0816s; samplesPerSecond = 14256.6
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.98228580 * 1085; EvalClassificationError = 0.52718894 * 1085; time = 0.0712s; samplesPerSecond = 15246.5
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.15 seconds since last report (0.01 seconds on comm.); 4826 samples processed by 2 workers (2249 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 33.08k samplesPerSecond , throughputPerWorker = 16.54k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.92865528 * 1164; EvalErrorPrediction = 0.54725086 * 1164; time = 0.0737s; samplesPerSecond = 15788.6
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 2.00184241 * 1167; EvalErrorPrediction = 0.57155099 * 1167; time = 0.0787s; samplesPerSecond = 14830.0
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.92865528 * 1164; EvalClassificationError = 0.54725086 * 1164; time = 0.0737s; samplesPerSecond = 15788.6
|
||||
MPI Rank 1: 08/16/2016 03:20:12: Epoch[ 3 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 2.00184241 * 1167; EvalClassificationError = 0.57155099 * 1167; time = 0.0787s; samplesPerSecond = 14830.0
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.15 seconds since last report (0.02 seconds on comm.); 4903 samples processed by 2 workers (2326 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 32.79k samplesPerSecond , throughputPerWorker = 16.39k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.92549525 * 1159; EvalErrorPrediction = 0.54270923 * 1159; time = 0.0701s; samplesPerSecond = 16542.0
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.98011842 * 823; EvalErrorPrediction = 0.54921021 * 823; time = 0.0835s; samplesPerSecond = 9857.7
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.92549525 * 1159; EvalClassificationError = 0.54270923 * 1159; time = 0.0701s; samplesPerSecond = 16542.0
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 3 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.98011842 * 823; EvalClassificationError = 0.54921021 * 823; time = 0.0835s; samplesPerSecond = 9857.7
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.17 seconds since last report (0.07 seconds on comm.); 5866 samples processed by 2 workers (823 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 34.21k samplesPerSecond , throughputPerWorker = 17.10k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.95703393 * 20480; EvalErrorPrediction = 0.54541016 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.67636s
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Finished Epoch[ 3 of 5]: [Training] CrossEntropyWithSoftmax = 1.95703393 * 20480; EvalClassificationError = 0.54541016 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.67636s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.89485201 * 1149; EvalErrorPrediction = 0.53176675 * 1149; time = 0.0931s; samplesPerSecond = 12338.1
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.89485201 * 1149; EvalClassificationError = 0.53176675 * 1149; time = 0.0931s; samplesPerSecond = 12338.1
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.17 seconds since last report (0.01 seconds on comm.); 4901 samples processed by 2 workers (2351 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 29.68k samplesPerSecond , throughputPerWorker = 14.84k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94072250 * 1202; EvalErrorPrediction = 0.53826955 * 1202; time = 0.0660s; samplesPerSecond = 18218.5
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.90064937 * 1130; EvalErrorPrediction = 0.52300885 * 1130; time = 0.0666s; samplesPerSecond = 16970.3
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94072250 * 1202; EvalClassificationError = 0.53826955 * 1202; time = 0.0660s; samplesPerSecond = 18218.5
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.90064937 * 1130; EvalClassificationError = 0.52300885 * 1130; time = 0.0666s; samplesPerSecond = 16970.3
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.13 seconds since last report (0.00 seconds on comm.); 4836 samples processed by 2 workers (2317 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 38.20k samplesPerSecond , throughputPerWorker = 19.10k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.85577719 * 1187; EvalErrorPrediction = 0.50884583 * 1187; time = 0.0591s; samplesPerSecond = 20098.2
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.94097997 * 1202; EvalErrorPrediction = 0.54658902 * 1202; time = 0.0818s; samplesPerSecond = 14687.2
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.85577719 * 1187; EvalClassificationError = 0.50884583 * 1187; time = 0.0591s; samplesPerSecond = 20098.2
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.94097997 * 1202; EvalClassificationError = 0.54658902 * 1202; time = 0.0818s; samplesPerSecond = 14687.2
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.15 seconds since last report (0.01 seconds on comm.); 4952 samples processed by 2 workers (2401 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 33.32k samplesPerSecond , throughputPerWorker = 16.66k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.91747174 * 1199; EvalErrorPrediction = 0.54211843 * 1199; time = 0.0657s; samplesPerSecond = 18254.9
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.91766783 * 817; EvalErrorPrediction = 0.53243574 * 817; time = 0.0473s; samplesPerSecond = 17271.6
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 1.91747174 * 1199; EvalClassificationError = 0.54211843 * 1199; time = 0.0657s; samplesPerSecond = 18254.9
|
||||
MPI Rank 1: 08/16/2016 03:20:13: Epoch[ 4 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.91766783 * 817; EvalClassificationError = 0.53243574 * 817; time = 0.0473s; samplesPerSecond = 17271.6
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.02-seconds latency this time; accumulated time on sync point = 0.02 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.16 seconds since last report (0.07 seconds on comm.); 5791 samples processed by 2 workers (817 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 35.86k samplesPerSecond , throughputPerWorker = 17.93k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.90306770 * 20480; EvalErrorPrediction = 0.52641602 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=0.604332s
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Finished Epoch[ 4 of 5]: [Training] CrossEntropyWithSoftmax = 1.90306770 * 20480; EvalClassificationError = 0.52641602 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=0.604332s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Starting Epoch 5: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: Parallel training (2 workers) using BlockMomentumSGD with block momentum = 0.5000, block momentum time constant (per worker) = 2954.6394, block learning rate = 1.0000, block size per worker = 2048 samples, using Nesterov-style block momentum, resetting SGD momentum after sync.
|
||||
MPI Rank 1: minibatchiterator: epoch 4: frames [81920..102400] (first utterance at frame 81920), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Starting minibatch loop, distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.86643684 * 1209; EvalErrorPrediction = 0.50372208 * 1209; time = 0.0891s; samplesPerSecond = 13571.8
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 1- 3, 15.00%]: CrossEntropyWithSoftmax = 1.86643684 * 1209; EvalClassificationError = 0.50372208 * 1209; time = 0.0891s; samplesPerSecond = 13571.8
|
||||
MPI Rank 1: (model aggregation stats): 1-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: 0.17 seconds since last report (0.01 seconds on comm.); 4919 samples processed by 2 workers (2426 by me);
|
||||
MPI Rank 1: (model aggregation stats) 1-th sync: totalThroughput = 29.63k samplesPerSecond , throughputPerWorker = 14.82k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94031579 * 1217; EvalErrorPrediction = 0.53327855 * 1217; time = 0.0696s; samplesPerSecond = 17488.1
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.89794045 * 1206; EvalErrorPrediction = 0.51824212 * 1206; time = 0.0600s; samplesPerSecond = 20116.1
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 4- 6, 30.00%]: CrossEntropyWithSoftmax = 1.94031579 * 1217; EvalClassificationError = 0.53327855 * 1217; time = 0.0696s; samplesPerSecond = 17488.1
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 7- 9, 45.00%]: CrossEntropyWithSoftmax = 1.89794045 * 1206; EvalClassificationError = 0.51824212 * 1206; time = 0.0600s; samplesPerSecond = 20116.1
|
||||
MPI Rank 1: (model aggregation stats): 2-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: 0.14 seconds since last report (0.00 seconds on comm.); 4899 samples processed by 2 workers (2419 by me);
|
||||
MPI Rank 1: (model aggregation stats) 2-th sync: totalThroughput = 35.49k samplesPerSecond , throughputPerWorker = 17.75k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.96415395 * 1213; EvalErrorPrediction = 0.53833471 * 1213; time = 0.0770s; samplesPerSecond = 15745.7
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.84516499 * 1147; EvalErrorPrediction = 0.51264167 * 1147; time = 0.0532s; samplesPerSecond = 21569.1
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 10- 12, 60.00%]: CrossEntropyWithSoftmax = 1.96415395 * 1213; EvalClassificationError = 0.53833471 * 1213; time = 0.0770s; samplesPerSecond = 15745.7
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 13- 15, 75.00%]: CrossEntropyWithSoftmax = 1.84516499 * 1147; EvalClassificationError = 0.51264167 * 1147; time = 0.0532s; samplesPerSecond = 21569.1
|
||||
MPI Rank 1: (model aggregation stats): 3-th sync point was hit, introducing a 0.00-seconds latency this time; accumulated time on sync point = 0.00 seconds , average latency = 0.00 seconds
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: 0.13 seconds since last report (0.00 seconds on comm.); 4829 samples processed by 2 workers (2359 by me);
|
||||
MPI Rank 1: (model aggregation stats) 3-th sync: totalThroughput = 37.07k samplesPerSecond , throughputPerWorker = 18.54k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 2.02566421 * 1212; EvalErrorPrediction = 0.55280528 * 1212; time = 0.0762s; samplesPerSecond = 15902.8
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.89201183 * 809; EvalErrorPrediction = 0.50679852 * 809; time = 0.0378s; samplesPerSecond = 21404.4
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 16- 18, 90.00%]: CrossEntropyWithSoftmax = 2.02566421 * 1212; EvalClassificationError = 0.55280528 * 1212; time = 0.0762s; samplesPerSecond = 15902.8
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Epoch[ 5 of 5]-Minibatch[ 19- 21, 105.00%]: CrossEntropyWithSoftmax = 1.89201183 * 809; EvalClassificationError = 0.50679852 * 809; time = 0.0378s; samplesPerSecond = 21404.4
|
||||
MPI Rank 1: (model aggregation stats): 4-th sync point was hit, introducing a 0.03-seconds latency this time; accumulated time on sync point = 0.03 seconds , average latency = 0.01 seconds
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: 0.17 seconds since last report (0.07 seconds on comm.); 5833 samples processed by 2 workers (809 by me);
|
||||
MPI Rank 1: (model aggregation stats) 4-th sync: totalThroughput = 34.97k samplesPerSecond , throughputPerWorker = 17.49k samplesPerSecond
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.88963745 * 20480; EvalErrorPrediction = 0.51865234 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-005; epochTime=0.603086s
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Finished Epoch[ 5 of 5]: [Training] CrossEntropyWithSoftmax = 1.88963745 * 20480; EvalClassificationError = 0.51865234 * 20480; totalSamplesSeen = 102400; learningRatePerSample = 9.7656251e-005; epochTime=0.603086s
|
||||
MPI Rank 1: 08/16/2016 03:20:14: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:14: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -133,7 +133,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -159,7 +159,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -242,7 +242,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -268,7 +268,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -357,7 +357,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -383,7 +383,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -479,7 +479,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -508,7 +508,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -531,7 +531,7 @@ MPI Rank 0: 08/16/2016 10:09:04: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 10:09:04: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:04: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 10:09:04: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 10:09:04: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -584,39 +584,39 @@ MPI Rank 0: 08/16/2016 10:09:07: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:07: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.3328s; samplesPerSecond = 1923.3
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.2984s; samplesPerSecond = 2145.0
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.1009s; samplesPerSecond = 6344.5
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1716s; samplesPerSecond = 3729.5
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.1670s; samplesPerSecond = 3832.5
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.1210s; samplesPerSecond = 5287.7
|
||||
MPI Rank 0: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.1979s; samplesPerSecond = 3233.3
|
||||
MPI Rank 0: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.2002s; samplesPerSecond = 3196.3
|
||||
MPI Rank 0: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.4442s; samplesPerSecond = 1440.6
|
||||
MPI Rank 0: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.2001s; samplesPerSecond = 3198.0
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.2024s; samplesPerSecond = 3162.3
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.1113s; samplesPerSecond = 5748.3
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.1265s; samplesPerSecond = 5058.8
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.2564s; samplesPerSecond = 2495.8
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.1036s; samplesPerSecond = 6180.5
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.0964s; samplesPerSecond = 6638.7
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.0944s; samplesPerSecond = 6780.7
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.1503s; samplesPerSecond = 4259.4
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.1362s; samplesPerSecond = 4700.2
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.0975s; samplesPerSecond = 6567.2
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1054s; samplesPerSecond = 6069.7
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.2378s; samplesPerSecond = 2691.6
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.1689s; samplesPerSecond = 3788.7
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.1497s; samplesPerSecond = 4274.0
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.1332s; samplesPerSecond = 4805.2
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.0999s; samplesPerSecond = 6404.8
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.1048s; samplesPerSecond = 6106.6
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1483s; samplesPerSecond = 4315.5
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.1339s; samplesPerSecond = 4779.7
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0906s; samplesPerSecond = 7064.7
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.0908s; samplesPerSecond = 7050.9
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.1003s; samplesPerSecond = 6380.7
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.1785s
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.3328s; samplesPerSecond = 1923.3
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.2984s; samplesPerSecond = 2145.0
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.1009s; samplesPerSecond = 6344.5
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1716s; samplesPerSecond = 3729.5
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.1670s; samplesPerSecond = 3832.5
|
||||
MPI Rank 0: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.1210s; samplesPerSecond = 5287.7
|
||||
MPI Rank 0: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.1979s; samplesPerSecond = 3233.3
|
||||
MPI Rank 0: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.2002s; samplesPerSecond = 3196.3
|
||||
MPI Rank 0: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.4442s; samplesPerSecond = 1440.6
|
||||
MPI Rank 0: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.2001s; samplesPerSecond = 3198.0
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.2024s; samplesPerSecond = 3162.3
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.1113s; samplesPerSecond = 5748.3
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.1265s; samplesPerSecond = 5058.8
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.2564s; samplesPerSecond = 2495.8
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.1036s; samplesPerSecond = 6180.5
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.0964s; samplesPerSecond = 6638.7
|
||||
MPI Rank 0: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.0944s; samplesPerSecond = 6780.7
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.1503s; samplesPerSecond = 4259.4
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.1362s; samplesPerSecond = 4700.2
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.0975s; samplesPerSecond = 6567.2
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1054s; samplesPerSecond = 6069.7
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.2378s; samplesPerSecond = 2691.6
|
||||
MPI Rank 0: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.1689s; samplesPerSecond = 3788.7
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.1497s; samplesPerSecond = 4274.0
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.1332s; samplesPerSecond = 4805.2
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.0999s; samplesPerSecond = 6404.8
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.1048s; samplesPerSecond = 6106.6
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1483s; samplesPerSecond = 4315.5
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.1339s; samplesPerSecond = 4779.7
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0906s; samplesPerSecond = 7064.7
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.0908s; samplesPerSecond = 7050.9
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.1003s; samplesPerSecond = 6380.7
|
||||
MPI Rank 0: 08/16/2016 10:09:12: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.1785s
|
||||
MPI Rank 0: 08/16/2016 10:09:14: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:14: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -626,45 +626,45 @@ MPI Rank 0: 08/16/2016 10:09:14: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.080039
|
||||
MPI Rank 0: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.025201
|
||||
MPI Rank 0: 08/16/2016 10:09:14: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.20516900 * 2304; EvalErrorPrediction = 0.61067708 * 2304; time = 0.4886s; samplesPerSecond = 4715.6
|
||||
MPI Rank 0: 08/16/2016 10:09:14: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.20516900 * 2304; EvalClassificationError = 0.61067708 * 2304; time = 0.4886s; samplesPerSecond = 4715.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.022399
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.025677
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1.1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.048238
|
||||
MPI Rank 0: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21634815 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.6421s; samplesPerSecond = 3986.9
|
||||
MPI Rank 0: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21634815 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.6421s; samplesPerSecond = 3986.9
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.022765
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.063737
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001326
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.022547
|
||||
MPI Rank 0: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.14548801 * 2560; EvalErrorPrediction = 0.58125000 * 2560; time = 0.5480s; samplesPerSecond = 4671.5
|
||||
MPI Rank 0: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.14548801 * 2560; EvalClassificationError = 0.58125000 * 2560; time = 0.5480s; samplesPerSecond = 4671.5
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.006938
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.013404
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.017959
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021878
|
||||
MPI Rank 0: 08/16/2016 10:09:16: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.14259959 * 2560; EvalErrorPrediction = 0.59414062 * 2560; time = 0.6469s; samplesPerSecond = 3957.2
|
||||
MPI Rank 0: 08/16/2016 10:09:16: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.14259959 * 2560; EvalClassificationError = 0.59414062 * 2560; time = 0.6469s; samplesPerSecond = 3957.2
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.056416
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.027043
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.073762
|
||||
MPI Rank 0: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18373009 * 2560; EvalErrorPrediction = 0.59375000 * 2560; time = 0.6071s; samplesPerSecond = 4216.6
|
||||
MPI Rank 0: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18373009 * 2560; EvalClassificationError = 0.59375000 * 2560; time = 0.6071s; samplesPerSecond = 4216.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.06793
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.067977
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.052426
|
||||
MPI Rank 0: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04918666 * 2560; EvalErrorPrediction = 0.56093750 * 2560; time = 0.6804s; samplesPerSecond = 3762.7
|
||||
MPI Rank 0: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04918666 * 2560; EvalClassificationError = 0.56093750 * 2560; time = 0.6804s; samplesPerSecond = 3762.7
|
||||
MPI Rank 0: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.023921
|
||||
MPI Rank 0: Async gradient aggregation wait time: 5e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.052635
|
||||
MPI Rank 0: 08/16/2016 10:09:18: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11591504 * 2560; EvalErrorPrediction = 0.61796875 * 2560; time = 0.7398s; samplesPerSecond = 3460.5
|
||||
MPI Rank 0: 08/16/2016 10:09:18: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11591504 * 2560; EvalClassificationError = 0.61796875 * 2560; time = 0.7398s; samplesPerSecond = 3460.5
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.008487
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.044247
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.051593
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.071231
|
||||
MPI Rank 0: 08/16/2016 10:09:19: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06399539 * 2560; EvalErrorPrediction = 0.57812500 * 2560; time = 0.5742s; samplesPerSecond = 4458.1
|
||||
MPI Rank 0: 08/16/2016 10:09:19: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06399539 * 2560; EvalClassificationError = 0.57812500 * 2560; time = 0.5742s; samplesPerSecond = 4458.1
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.039068
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.074004
|
||||
MPI Rank 0: 08/16/2016 10:09:19: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14072163 * 20480; EvalErrorPrediction = 0.58969727 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=5.04684s
|
||||
MPI Rank 0: 08/16/2016 10:09:19: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14072163 * 20480; EvalClassificationError = 0.58969727 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=5.04684s
|
||||
MPI Rank 0: 08/16/2016 10:09:19: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:19: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -675,13 +675,13 @@ MPI Rank 0: Async gradient aggregation wait time: 0.011011
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.088497
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.026596
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.074608
|
||||
MPI Rank 0: 08/16/2016 10:09:20: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.19691730 * 9216; EvalErrorPrediction = 0.58257378 * 9216; time = 0.8724s; samplesPerSecond = 10564.4
|
||||
MPI Rank 0: 08/16/2016 10:09:20: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.19691730 * 9216; EvalClassificationError = 0.58257378 * 9216; time = 0.8724s; samplesPerSecond = 10564.4
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.127009
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.106847
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.063592
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.064724
|
||||
MPI Rank 0: 08/16/2016 10:09:21: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.11412321 * 10240; EvalErrorPrediction = 0.57666016 * 10240; time = 1.0823s; samplesPerSecond = 9461.1
|
||||
MPI Rank 0: 08/16/2016 10:09:21: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.14813003 * 20480; EvalErrorPrediction = 0.57880859 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.31594s
|
||||
MPI Rank 0: 08/16/2016 10:09:21: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.11412321 * 10240; EvalClassificationError = 0.57666016 * 10240; time = 1.0823s; samplesPerSecond = 9461.1
|
||||
MPI Rank 0: 08/16/2016 10:09:21: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.14813003 * 20480; EvalClassificationError = 0.57880859 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.31594s
|
||||
MPI Rank 0: 08/16/2016 10:09:21: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:21: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -692,14 +692,14 @@ MPI Rank 0: Async gradient aggregation wait time: 1.1e-05
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.023009
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.020995
|
||||
MPI Rank 0: 08/16/2016 10:09:23: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97540136 * 9216; EvalErrorPrediction = 0.53862847 * 9216; time = 1.6293s; samplesPerSecond = 5656.5
|
||||
MPI Rank 0: 08/16/2016 10:09:23: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97540136 * 9216; EvalClassificationError = 0.53862847 * 9216; time = 1.6293s; samplesPerSecond = 5656.5
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1.1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.054934
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.036059
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.121828
|
||||
MPI Rank 0: 08/16/2016 10:09:24: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95172037 * 10240; EvalErrorPrediction = 0.53369141 * 10240; time = 1.2943s; samplesPerSecond = 7911.9
|
||||
MPI Rank 0: 08/16/2016 10:09:24: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95172037 * 10240; EvalClassificationError = 0.53369141 * 10240; time = 1.2943s; samplesPerSecond = 7911.9
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.07567
|
||||
MPI Rank 0: 08/16/2016 10:09:24: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.96347835 * 20480; EvalErrorPrediction = 0.53618164 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=3.0298s
|
||||
MPI Rank 0: 08/16/2016 10:09:24: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.96347835 * 20480; EvalClassificationError = 0.53618164 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=3.0298s
|
||||
MPI Rank 0: 08/16/2016 10:09:24: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 10:09:24: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -752,7 +752,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -778,7 +778,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -861,7 +861,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -887,7 +887,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -976,7 +976,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1002,7 +1002,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1098,7 +1098,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1127,7 +1127,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1150,7 +1150,7 @@ MPI Rank 1: 08/16/2016 10:09:04: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 10:09:04: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:04: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 10:09:04: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 10:09:04: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1203,39 +1203,39 @@ MPI Rank 1: 08/16/2016 10:09:07: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:07: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 10:09:07: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.0667s; samplesPerSecond = 9594.8
|
||||
MPI Rank 1: 08/16/2016 10:09:07: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.0537s; samplesPerSecond = 11921.2
|
||||
MPI Rank 1: 08/16/2016 10:09:07: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0540s; samplesPerSecond = 11854.9
|
||||
MPI Rank 1: 08/16/2016 10:09:07: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0539s; samplesPerSecond = 11876.5
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.0540s; samplesPerSecond = 11844.0
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.0539s; samplesPerSecond = 11866.8
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.0546s; samplesPerSecond = 11713.9
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.0549s; samplesPerSecond = 11662.2
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.0539s; samplesPerSecond = 11865.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.0537s; samplesPerSecond = 11926.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.0537s; samplesPerSecond = 11915.4
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.0540s; samplesPerSecond = 11858.0
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0538s; samplesPerSecond = 11900.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.0541s; samplesPerSecond = 11838.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.0558s; samplesPerSecond = 11471.4
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.0540s; samplesPerSecond = 11860.0
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.0540s; samplesPerSecond = 11860.9
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.0540s; samplesPerSecond = 11842.2
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.0541s; samplesPerSecond = 11836.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.0541s; samplesPerSecond = 11829.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0540s; samplesPerSecond = 11856.9
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.0542s; samplesPerSecond = 11801.8
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0545s; samplesPerSecond = 11733.6
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0545s; samplesPerSecond = 11738.8
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0526s; samplesPerSecond = 12169.2
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.0527s; samplesPerSecond = 12144.2
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.0533s; samplesPerSecond = 12003.5
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0527s; samplesPerSecond = 12150.0
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.0528s; samplesPerSecond = 12124.9
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0520s; samplesPerSecond = 12302.3
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.0545s; samplesPerSecond = 11738.8
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.0524s; samplesPerSecond = 12217.0
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.74073s
|
||||
MPI Rank 1: 08/16/2016 10:09:07: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.0667s; samplesPerSecond = 9594.8
|
||||
MPI Rank 1: 08/16/2016 10:09:07: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.0537s; samplesPerSecond = 11921.2
|
||||
MPI Rank 1: 08/16/2016 10:09:07: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0540s; samplesPerSecond = 11854.9
|
||||
MPI Rank 1: 08/16/2016 10:09:07: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0539s; samplesPerSecond = 11876.5
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.0540s; samplesPerSecond = 11844.0
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.0539s; samplesPerSecond = 11866.8
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.0546s; samplesPerSecond = 11713.9
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.0549s; samplesPerSecond = 11662.2
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.0539s; samplesPerSecond = 11865.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.0537s; samplesPerSecond = 11926.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.0537s; samplesPerSecond = 11915.4
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.0540s; samplesPerSecond = 11858.0
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0538s; samplesPerSecond = 11900.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.0541s; samplesPerSecond = 11838.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.0558s; samplesPerSecond = 11471.4
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.0540s; samplesPerSecond = 11860.0
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.0540s; samplesPerSecond = 11860.9
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.0540s; samplesPerSecond = 11842.2
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.0541s; samplesPerSecond = 11836.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.0541s; samplesPerSecond = 11829.3
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0540s; samplesPerSecond = 11856.9
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.0542s; samplesPerSecond = 11801.8
|
||||
MPI Rank 1: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0545s; samplesPerSecond = 11733.6
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0545s; samplesPerSecond = 11738.8
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0526s; samplesPerSecond = 12169.2
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.0527s; samplesPerSecond = 12144.2
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.0533s; samplesPerSecond = 12003.5
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0527s; samplesPerSecond = 12150.0
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.0528s; samplesPerSecond = 12124.9
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0520s; samplesPerSecond = 12302.3
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.0545s; samplesPerSecond = 11738.8
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.0524s; samplesPerSecond = 12217.0
|
||||
MPI Rank 1: 08/16/2016 10:09:09: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.74073s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:14: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1244,45 +1244,45 @@ MPI Rank 1: 08/16/2016 10:09:14: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.092054
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.029108
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.053094
|
||||
MPI Rank 1: 08/16/2016 10:09:14: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.20516900 * 2304; EvalErrorPrediction = 0.61067708 * 2304; time = 0.4871s; samplesPerSecond = 4730.2
|
||||
MPI Rank 1: 08/16/2016 10:09:14: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.20516900 * 2304; EvalClassificationError = 0.61067708 * 2304; time = 0.4871s; samplesPerSecond = 4730.2
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.029805
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.034276
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.035131
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.063289
|
||||
MPI Rank 1: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21634815 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.6433s; samplesPerSecond = 3979.4
|
||||
MPI Rank 1: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21634815 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.6433s; samplesPerSecond = 3979.4
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.033417
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.048684
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.052807
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.032373
|
||||
MPI Rank 1: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.14548801 * 2560; EvalErrorPrediction = 0.58125000 * 2560; time = 0.5381s; samplesPerSecond = 4757.3
|
||||
MPI Rank 1: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.14548801 * 2560; EvalClassificationError = 0.58125000 * 2560; time = 0.5381s; samplesPerSecond = 4757.3
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.047659
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.025607
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.199686
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.057808
|
||||
MPI Rank 1: 08/16/2016 10:09:16: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.14259959 * 2560; EvalErrorPrediction = 0.59414062 * 2560; time = 0.6424s; samplesPerSecond = 3985.1
|
||||
MPI Rank 1: 08/16/2016 10:09:16: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.14259959 * 2560; EvalClassificationError = 0.59414062 * 2560; time = 0.6424s; samplesPerSecond = 3985.1
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.037252
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.064611
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.04896
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.063044
|
||||
MPI Rank 1: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18373009 * 2560; EvalErrorPrediction = 0.59375000 * 2560; time = 0.5712s; samplesPerSecond = 4482.0
|
||||
MPI Rank 1: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18373009 * 2560; EvalClassificationError = 0.59375000 * 2560; time = 0.5712s; samplesPerSecond = 4482.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.17443
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.050119
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.055181
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.068994
|
||||
MPI Rank 1: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04918666 * 2560; EvalErrorPrediction = 0.56093750 * 2560; time = 0.7096s; samplesPerSecond = 3607.7
|
||||
MPI Rank 1: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04918666 * 2560; EvalClassificationError = 0.56093750 * 2560; time = 0.7096s; samplesPerSecond = 3607.7
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.042221
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.058501
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.055313
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.05451
|
||||
MPI Rank 1: 08/16/2016 10:09:18: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11591504 * 2560; EvalErrorPrediction = 0.61796875 * 2560; time = 0.7398s; samplesPerSecond = 3460.2
|
||||
MPI Rank 1: 08/16/2016 10:09:18: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11591504 * 2560; EvalClassificationError = 0.61796875 * 2560; time = 0.7398s; samplesPerSecond = 3460.2
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.038832
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.05246
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.057669
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.092341
|
||||
MPI Rank 1: 08/16/2016 10:09:19: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06399539 * 2560; EvalErrorPrediction = 0.57812500 * 2560; time = 0.5680s; samplesPerSecond = 4507.3
|
||||
MPI Rank 1: 08/16/2016 10:09:19: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06399539 * 2560; EvalClassificationError = 0.57812500 * 2560; time = 0.5680s; samplesPerSecond = 4507.3
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.062751
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.053938
|
||||
MPI Rank 1: 08/16/2016 10:09:19: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14072163 * 20480; EvalErrorPrediction = 0.58969727 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=5.01863s
|
||||
MPI Rank 1: 08/16/2016 10:09:19: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14072163 * 20480; EvalClassificationError = 0.58969727 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=5.01863s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:19: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1292,13 +1292,13 @@ MPI Rank 1: Async gradient aggregation wait time: 0.009871
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.084551
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.067075
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.088626
|
||||
MPI Rank 1: 08/16/2016 10:09:20: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.19691730 * 9216; EvalErrorPrediction = 0.58257378 * 9216; time = 0.8646s; samplesPerSecond = 10659.1
|
||||
MPI Rank 1: 08/16/2016 10:09:20: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.19691730 * 9216; EvalClassificationError = 0.58257378 * 9216; time = 0.8646s; samplesPerSecond = 10659.1
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.165435
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.107027
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.079878
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.071622
|
||||
MPI Rank 1: 08/16/2016 10:09:21: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.11412321 * 10240; EvalErrorPrediction = 0.57666016 * 10240; time = 1.0905s; samplesPerSecond = 9390.4
|
||||
MPI Rank 1: 08/16/2016 10:09:21: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.14813003 * 20480; EvalErrorPrediction = 0.57880859 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.30731s
|
||||
MPI Rank 1: 08/16/2016 10:09:21: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.11412321 * 10240; EvalClassificationError = 0.57666016 * 10240; time = 1.0905s; samplesPerSecond = 9390.4
|
||||
MPI Rank 1: 08/16/2016 10:09:21: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.14813003 * 20480; EvalClassificationError = 0.57880859 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.30731s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:21: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1308,14 +1308,14 @@ MPI Rank 1: Async gradient aggregation wait time: 0.162303
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.088365
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.357011
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.142726
|
||||
MPI Rank 1: 08/16/2016 10:09:23: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97540136 * 9216; EvalErrorPrediction = 0.53862847 * 9216; time = 1.5273s; samplesPerSecond = 6034.1
|
||||
MPI Rank 1: 08/16/2016 10:09:23: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97540136 * 9216; EvalClassificationError = 0.53862847 * 9216; time = 1.5273s; samplesPerSecond = 6034.1
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.086634
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.244028
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.136427
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.134996
|
||||
MPI Rank 1: 08/16/2016 10:09:24: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95172037 * 10240; EvalErrorPrediction = 0.53369141 * 10240; time = 1.3349s; samplesPerSecond = 7670.9
|
||||
MPI Rank 1: 08/16/2016 10:09:24: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95172037 * 10240; EvalClassificationError = 0.53369141 * 10240; time = 1.3349s; samplesPerSecond = 7670.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.059731
|
||||
MPI Rank 1: 08/16/2016 10:09:24: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.96347835 * 20480; EvalErrorPrediction = 0.53618164 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=3.01163s
|
||||
MPI Rank 1: 08/16/2016 10:09:24: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.96347835 * 20480; EvalClassificationError = 0.53618164 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=3.01163s
|
||||
MPI Rank 1: 08/16/2016 10:09:24: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:24: Action "train" complete.
|
||||
|
|
@ -1367,7 +1367,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1393,7 +1393,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1476,7 +1476,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1502,7 +1502,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1591,7 +1591,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1617,7 +1617,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1713,7 +1713,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1742,7 +1742,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1765,7 +1765,7 @@ MPI Rank 2: 08/16/2016 10:09:05: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 10:09:05: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:05: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 10:09:05: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 10:09:05: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1818,39 +1818,39 @@ MPI Rank 2: 08/16/2016 10:09:07: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:07: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 10:09:07: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.2390s; samplesPerSecond = 2678.0
|
||||
MPI Rank 2: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.5079s; samplesPerSecond = 1260.0
|
||||
MPI Rank 2: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.2649s; samplesPerSecond = 2415.8
|
||||
MPI Rank 2: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.2718s; samplesPerSecond = 2354.7
|
||||
MPI Rank 2: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.2604s; samplesPerSecond = 2457.8
|
||||
MPI Rank 2: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.3544s; samplesPerSecond = 1806.0
|
||||
MPI Rank 2: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.2613s; samplesPerSecond = 2449.7
|
||||
MPI Rank 2: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.2489s; samplesPerSecond = 2571.6
|
||||
MPI Rank 2: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.4933s; samplesPerSecond = 1297.4
|
||||
MPI Rank 2: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.3388s; samplesPerSecond = 1888.9
|
||||
MPI Rank 2: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.2674s; samplesPerSecond = 2393.0
|
||||
MPI Rank 2: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.3065s; samplesPerSecond = 2087.9
|
||||
MPI Rank 2: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.4169s; samplesPerSecond = 1535.2
|
||||
MPI Rank 2: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.3248s; samplesPerSecond = 1970.4
|
||||
MPI Rank 2: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.5251s; samplesPerSecond = 1218.8
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.3305s; samplesPerSecond = 1936.7
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.0838s; samplesPerSecond = 7639.2
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.0547s; samplesPerSecond = 11707.0
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.0548s; samplesPerSecond = 11674.6
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.0548s; samplesPerSecond = 11673.3
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0544s; samplesPerSecond = 11762.8
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.0549s; samplesPerSecond = 11661.0
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0546s; samplesPerSecond = 11726.3
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0547s; samplesPerSecond = 11700.2
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0547s; samplesPerSecond = 11705.7
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.0548s; samplesPerSecond = 11676.5
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.0546s; samplesPerSecond = 11716.2
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0544s; samplesPerSecond = 11762.5
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.0540s; samplesPerSecond = 11841.8
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0532s; samplesPerSecond = 12023.1
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.0522s; samplesPerSecond = 12262.4
|
||||
MPI Rank 2: 08/16/2016 10:09:14: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.0527s; samplesPerSecond = 12147.2
|
||||
MPI Rank 2: 08/16/2016 10:09:14: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=6.31457s
|
||||
MPI Rank 2: 08/16/2016 10:09:07: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.2390s; samplesPerSecond = 2678.0
|
||||
MPI Rank 2: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.5079s; samplesPerSecond = 1260.0
|
||||
MPI Rank 2: 08/16/2016 10:09:08: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.2649s; samplesPerSecond = 2415.8
|
||||
MPI Rank 2: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.2718s; samplesPerSecond = 2354.7
|
||||
MPI Rank 2: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.2604s; samplesPerSecond = 2457.8
|
||||
MPI Rank 2: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.3544s; samplesPerSecond = 1806.0
|
||||
MPI Rank 2: 08/16/2016 10:09:09: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.2613s; samplesPerSecond = 2449.7
|
||||
MPI Rank 2: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.2489s; samplesPerSecond = 2571.6
|
||||
MPI Rank 2: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.4933s; samplesPerSecond = 1297.4
|
||||
MPI Rank 2: 08/16/2016 10:09:10: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.3388s; samplesPerSecond = 1888.9
|
||||
MPI Rank 2: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.2674s; samplesPerSecond = 2393.0
|
||||
MPI Rank 2: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.3065s; samplesPerSecond = 2087.9
|
||||
MPI Rank 2: 08/16/2016 10:09:11: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.4169s; samplesPerSecond = 1535.2
|
||||
MPI Rank 2: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.3248s; samplesPerSecond = 1970.4
|
||||
MPI Rank 2: 08/16/2016 10:09:12: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.5251s; samplesPerSecond = 1218.8
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.3305s; samplesPerSecond = 1936.7
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.0838s; samplesPerSecond = 7639.2
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.0547s; samplesPerSecond = 11707.0
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.0548s; samplesPerSecond = 11674.6
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.0548s; samplesPerSecond = 11673.3
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0544s; samplesPerSecond = 11762.8
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.0549s; samplesPerSecond = 11661.0
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0546s; samplesPerSecond = 11726.3
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0547s; samplesPerSecond = 11700.2
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0547s; samplesPerSecond = 11705.7
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.0548s; samplesPerSecond = 11676.5
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.0546s; samplesPerSecond = 11716.2
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0544s; samplesPerSecond = 11762.5
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.0540s; samplesPerSecond = 11841.8
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0532s; samplesPerSecond = 12023.1
|
||||
MPI Rank 2: 08/16/2016 10:09:13: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.0522s; samplesPerSecond = 12262.4
|
||||
MPI Rank 2: 08/16/2016 10:09:14: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.0527s; samplesPerSecond = 12147.2
|
||||
MPI Rank 2: 08/16/2016 10:09:14: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=6.31457s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:14: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1859,45 +1859,45 @@ MPI Rank 2: 08/16/2016 10:09:14: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.124401
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.027767
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.053848
|
||||
MPI Rank 2: 08/16/2016 10:09:14: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.20516900 * 2304; EvalErrorPrediction = 0.61067708 * 2304; time = 0.4975s; samplesPerSecond = 4630.8
|
||||
MPI Rank 2: 08/16/2016 10:09:14: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.20516900 * 2304; EvalClassificationError = 0.61067708 * 2304; time = 0.4975s; samplesPerSecond = 4630.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.007784
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.021951
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.04167
|
||||
MPI Rank 2: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21634815 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.6501s; samplesPerSecond = 3937.6
|
||||
MPI Rank 2: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21634815 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.6501s; samplesPerSecond = 3937.6
|
||||
MPI Rank 2: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.054872
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.026931
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.03089
|
||||
MPI Rank 2: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.14548801 * 2560; EvalErrorPrediction = 0.58125000 * 2560; time = 0.5396s; samplesPerSecond = 4744.7
|
||||
MPI Rank 2: 08/16/2016 10:09:15: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.14548801 * 2560; EvalClassificationError = 0.58125000 * 2560; time = 0.5396s; samplesPerSecond = 4744.7
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.014629
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.02573
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.116871
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.055267
|
||||
MPI Rank 2: 08/16/2016 10:09:16: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.14259959 * 2560; EvalErrorPrediction = 0.59414062 * 2560; time = 0.6542s; samplesPerSecond = 3912.9
|
||||
MPI Rank 2: 08/16/2016 10:09:16: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.14259959 * 2560; EvalClassificationError = 0.59414062 * 2560; time = 0.6542s; samplesPerSecond = 3912.9
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1.4e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.066885
|
||||
MPI Rank 2: Async gradient aggregation wait time: 8e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.053743
|
||||
MPI Rank 2: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18373009 * 2560; EvalErrorPrediction = 0.59375000 * 2560; time = 0.6964s; samplesPerSecond = 3675.8
|
||||
MPI Rank 2: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18373009 * 2560; EvalClassificationError = 0.59375000 * 2560; time = 0.6964s; samplesPerSecond = 3675.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.045098
|
||||
MPI Rank 2: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.034369
|
||||
MPI Rank 2: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04918666 * 2560; EvalErrorPrediction = 0.56093750 * 2560; time = 0.5675s; samplesPerSecond = 4510.7
|
||||
MPI Rank 2: 08/16/2016 10:09:17: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04918666 * 2560; EvalClassificationError = 0.56093750 * 2560; time = 0.5675s; samplesPerSecond = 4510.7
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.011305
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.056924
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.026861
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.050577
|
||||
MPI Rank 2: 08/16/2016 10:09:18: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11591504 * 2560; EvalErrorPrediction = 0.61796875 * 2560; time = 0.7274s; samplesPerSecond = 3519.2
|
||||
MPI Rank 2: 08/16/2016 10:09:18: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11591504 * 2560; EvalClassificationError = 0.61796875 * 2560; time = 0.7274s; samplesPerSecond = 3519.2
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.005283
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.051956
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.054077
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.04628
|
||||
MPI Rank 2: 08/16/2016 10:09:19: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06399539 * 2560; EvalErrorPrediction = 0.57812500 * 2560; time = 0.5709s; samplesPerSecond = 4483.9
|
||||
MPI Rank 2: 08/16/2016 10:09:19: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06399539 * 2560; EvalClassificationError = 0.57812500 * 2560; time = 0.5709s; samplesPerSecond = 4483.9
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.062883
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.064039
|
||||
MPI Rank 2: 08/16/2016 10:09:19: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14072163 * 20480; EvalErrorPrediction = 0.58969727 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=5.04153s
|
||||
MPI Rank 2: 08/16/2016 10:09:19: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14072163 * 20480; EvalClassificationError = 0.58969727 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=5.04153s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:19: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1907,13 +1907,13 @@ MPI Rank 2: Async gradient aggregation wait time: 1.1e-05
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.034828
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1.1e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.025138
|
||||
MPI Rank 2: 08/16/2016 10:09:20: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.19691730 * 9216; EvalErrorPrediction = 0.58257378 * 9216; time = 0.9090s; samplesPerSecond = 10138.8
|
||||
MPI Rank 2: 08/16/2016 10:09:20: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.19691730 * 9216; EvalClassificationError = 0.58257378 * 9216; time = 0.9090s; samplesPerSecond = 10138.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.022031
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1.1e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.013082
|
||||
MPI Rank 2: 08/16/2016 10:09:21: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.11412321 * 10240; EvalErrorPrediction = 0.57666016 * 10240; time = 1.3002s; samplesPerSecond = 7875.9
|
||||
MPI Rank 2: 08/16/2016 10:09:21: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.14813003 * 20480; EvalErrorPrediction = 0.57880859 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.31728s
|
||||
MPI Rank 2: 08/16/2016 10:09:21: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.11412321 * 10240; EvalClassificationError = 0.57666016 * 10240; time = 1.3002s; samplesPerSecond = 7875.9
|
||||
MPI Rank 2: 08/16/2016 10:09:21: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.14813003 * 20480; EvalClassificationError = 0.57880859 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.31728s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:21: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1923,14 +1923,14 @@ MPI Rank 2: Async gradient aggregation wait time: 0.144867
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.087324
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.337574
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.132562
|
||||
MPI Rank 2: 08/16/2016 10:09:23: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97540136 * 9216; EvalErrorPrediction = 0.53862847 * 9216; time = 1.5385s; samplesPerSecond = 5990.2
|
||||
MPI Rank 2: 08/16/2016 10:09:23: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97540136 * 9216; EvalClassificationError = 0.53862847 * 9216; time = 1.5385s; samplesPerSecond = 5990.2
|
||||
MPI Rank 2: Async gradient aggregation wait time: 7e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.256707
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1.2e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.022902
|
||||
MPI Rank 2: 08/16/2016 10:09:24: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95172037 * 10240; EvalErrorPrediction = 0.53369141 * 10240; time = 1.3683s; samplesPerSecond = 7483.7
|
||||
MPI Rank 2: 08/16/2016 10:09:24: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95172037 * 10240; EvalClassificationError = 0.53369141 * 10240; time = 1.3683s; samplesPerSecond = 7483.7
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.069458
|
||||
MPI Rank 2: 08/16/2016 10:09:24: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.96347835 * 20480; EvalErrorPrediction = 0.53618164 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=3.03183s
|
||||
MPI Rank 2: 08/16/2016 10:09:24: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.96347835 * 20480; EvalClassificationError = 0.53618164 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=3.03183s
|
||||
MPI Rank 2: 08/16/2016 10:09:24: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:24: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -133,7 +133,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -159,7 +159,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -242,7 +242,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -268,7 +268,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -357,7 +357,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -383,7 +383,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -480,7 +480,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -509,7 +509,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -532,7 +532,7 @@ MPI Rank 0: 08/16/2016 10:09:30: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 10:09:30: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:30: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 10:09:30: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 10:09:30: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -585,39 +585,39 @@ MPI Rank 0: 08/16/2016 10:09:34: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1000s; samplesPerSecond = 6401.1
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0985s; samplesPerSecond = 6496.0
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0985s; samplesPerSecond = 6496.5
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0986s; samplesPerSecond = 6488.3
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0986s; samplesPerSecond = 6487.7
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0987s; samplesPerSecond = 6486.3
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0987s; samplesPerSecond = 6487.3
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.0986s; samplesPerSecond = 6490.1
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0986s; samplesPerSecond = 6491.6
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0987s; samplesPerSecond = 6486.0
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0987s; samplesPerSecond = 6485.4
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0986s; samplesPerSecond = 6494.0
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0985s; samplesPerSecond = 6496.1
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0986s; samplesPerSecond = 6491.3
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6486.9
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0987s; samplesPerSecond = 6481.7
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0987s; samplesPerSecond = 6487.4
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0987s; samplesPerSecond = 6486.9
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0987s; samplesPerSecond = 6486.7
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0986s; samplesPerSecond = 6487.8
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0987s; samplesPerSecond = 6485.4
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6491.9
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6488.4
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0987s; samplesPerSecond = 6485.5
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0987s; samplesPerSecond = 6482.2
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0987s; samplesPerSecond = 6481.2
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0987s; samplesPerSecond = 6485.5
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0987s; samplesPerSecond = 6484.1
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0987s; samplesPerSecond = 6484.2
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0987s; samplesPerSecond = 6483.2
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0987s; samplesPerSecond = 6481.5
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0986s; samplesPerSecond = 6490.9
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.1617s
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1000s; samplesPerSecond = 6401.1
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0985s; samplesPerSecond = 6496.0
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0985s; samplesPerSecond = 6496.5
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0986s; samplesPerSecond = 6488.3
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0986s; samplesPerSecond = 6487.7
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0987s; samplesPerSecond = 6486.3
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0987s; samplesPerSecond = 6487.3
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.0986s; samplesPerSecond = 6490.1
|
||||
MPI Rank 0: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0986s; samplesPerSecond = 6491.6
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0987s; samplesPerSecond = 6486.0
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0987s; samplesPerSecond = 6485.4
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0986s; samplesPerSecond = 6494.0
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0985s; samplesPerSecond = 6496.1
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0986s; samplesPerSecond = 6491.3
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6486.9
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0987s; samplesPerSecond = 6481.7
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0987s; samplesPerSecond = 6487.4
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0987s; samplesPerSecond = 6486.9
|
||||
MPI Rank 0: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0987s; samplesPerSecond = 6486.7
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0986s; samplesPerSecond = 6487.8
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0987s; samplesPerSecond = 6485.4
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6491.9
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6488.4
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0987s; samplesPerSecond = 6485.5
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0987s; samplesPerSecond = 6482.2
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0987s; samplesPerSecond = 6481.2
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0987s; samplesPerSecond = 6485.5
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0987s; samplesPerSecond = 6484.1
|
||||
MPI Rank 0: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0987s; samplesPerSecond = 6484.2
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0987s; samplesPerSecond = 6483.2
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0987s; samplesPerSecond = 6481.5
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0986s; samplesPerSecond = 6490.9
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.1617s
|
||||
MPI Rank 0: 08/16/2016 10:09:37: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -627,45 +627,45 @@ MPI Rank 0: 08/16/2016 10:09:37: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.006881
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001169
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.012812
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalErrorPrediction = 0.61111111 * 2304; time = 0.1384s; samplesPerSecond = 16646.2
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalClassificationError = 0.61111111 * 2304; time = 0.1384s; samplesPerSecond = 16646.2
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.005635
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.014907
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.006234
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.1369s; samplesPerSecond = 18705.7
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.1369s; samplesPerSecond = 18705.7
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00159
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.013101
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001606
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.013122
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalErrorPrediction = 0.57617188 * 2560; time = 0.1323s; samplesPerSecond = 19348.9
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalClassificationError = 0.57617188 * 2560; time = 0.1323s; samplesPerSecond = 19348.9
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001655
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.011781
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001591
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.013005
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalErrorPrediction = 0.60664063 * 2560; time = 0.1335s; samplesPerSecond = 19172.6
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalClassificationError = 0.60664063 * 2560; time = 0.1335s; samplesPerSecond = 19172.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001611
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.013065
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001604
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.013188
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalErrorPrediction = 0.58945313 * 2560; time = 0.1329s; samplesPerSecond = 19262.9
|
||||
MPI Rank 0: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalClassificationError = 0.58945313 * 2560; time = 0.1329s; samplesPerSecond = 19262.9
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001593
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.013056
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.003171
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.01456
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalErrorPrediction = 0.56562500 * 2560; time = 0.1343s; samplesPerSecond = 19056.6
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalClassificationError = 0.56562500 * 2560; time = 0.1343s; samplesPerSecond = 19056.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001389
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.01173
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001655
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.011772
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalErrorPrediction = 0.59179688 * 2560; time = 0.1324s; samplesPerSecond = 19332.7
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalClassificationError = 0.59179688 * 2560; time = 0.1324s; samplesPerSecond = 19332.7
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001683
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.011663
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00557
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.013259
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalErrorPrediction = 0.58632812 * 2560; time = 0.1331s; samplesPerSecond = 19231.9
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalClassificationError = 0.58632812 * 2560; time = 0.1331s; samplesPerSecond = 19231.9
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00369
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.002638
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalErrorPrediction = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.08374s
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalClassificationError = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.08374s
|
||||
MPI Rank 0: 08/16/2016 10:09:38: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -676,13 +676,13 @@ MPI Rank 0: Async gradient aggregation wait time: 0.003256
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.026681
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001712
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.027434
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalErrorPrediction = 0.56510417 * 9216; time = 0.2771s; samplesPerSecond = 33259.5
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalClassificationError = 0.56510417 * 9216; time = 0.2771s; samplesPerSecond = 33259.5
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001193
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.026669
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002323
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.022121
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalErrorPrediction = 0.56914062 * 10240; time = 0.2697s; samplesPerSecond = 37967.3
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalErrorPrediction = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.563103s
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalClassificationError = 0.56914062 * 10240; time = 0.2697s; samplesPerSecond = 37967.3
|
||||
MPI Rank 0: 08/16/2016 10:09:38: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalClassificationError = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.563103s
|
||||
MPI Rank 0: 08/16/2016 10:09:39: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:09:39: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -693,14 +693,14 @@ MPI Rank 0: Async gradient aggregation wait time: 0.001962
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.02659
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.003671
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.026639
|
||||
MPI Rank 0: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalErrorPrediction = 0.53982205 * 9216; time = 0.2710s; samplesPerSecond = 34010.8
|
||||
MPI Rank 0: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalClassificationError = 0.53982205 * 9216; time = 0.2710s; samplesPerSecond = 34010.8
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00183
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.026213
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.001929
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.026224
|
||||
MPI Rank 0: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalErrorPrediction = 0.53369141 * 10240; time = 0.2738s; samplesPerSecond = 37394.6
|
||||
MPI Rank 0: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalClassificationError = 0.53369141 * 10240; time = 0.2738s; samplesPerSecond = 37394.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.003482
|
||||
MPI Rank 0: 08/16/2016 10:09:39: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalErrorPrediction = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.555616s
|
||||
MPI Rank 0: 08/16/2016 10:09:39: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalClassificationError = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.555616s
|
||||
MPI Rank 0: 08/16/2016 10:09:39: SGD: Saving checkpoint model '/tmp/cntk-test-20160816100737.796285/Speech/DNN_ParallelBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 10:09:39: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -753,7 +753,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -779,7 +779,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -862,7 +862,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -888,7 +888,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -977,7 +977,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1003,7 +1003,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1100,7 +1100,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1129,7 +1129,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1152,7 +1152,7 @@ MPI Rank 1: 08/16/2016 10:09:29: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 10:09:29: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:29: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 10:09:29: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 10:09:29: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1205,39 +1205,39 @@ MPI Rank 1: 08/16/2016 10:09:34: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.0983s; samplesPerSecond = 6509.8
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0985s; samplesPerSecond = 6496.4
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0985s; samplesPerSecond = 6495.7
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0986s; samplesPerSecond = 6487.8
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0986s; samplesPerSecond = 6488.5
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0987s; samplesPerSecond = 6485.1
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0987s; samplesPerSecond = 6487.4
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.0986s; samplesPerSecond = 6490.1
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0986s; samplesPerSecond = 6491.3
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0987s; samplesPerSecond = 6486.6
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0987s; samplesPerSecond = 6485.0
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0986s; samplesPerSecond = 6493.4
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0985s; samplesPerSecond = 6495.7
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0986s; samplesPerSecond = 6491.1
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6487.2
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0988s; samplesPerSecond = 6480.5
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0987s; samplesPerSecond = 6487.1
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0987s; samplesPerSecond = 6487.6
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0987s; samplesPerSecond = 6485.7
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0987s; samplesPerSecond = 6487.3
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0987s; samplesPerSecond = 6483.9
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6491.6
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6488.6
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0987s; samplesPerSecond = 6485.3
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0987s; samplesPerSecond = 6482.3
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0988s; samplesPerSecond = 6481.0
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0987s; samplesPerSecond = 6485.5
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0987s; samplesPerSecond = 6484.0
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0987s; samplesPerSecond = 6482.3
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0987s; samplesPerSecond = 6484.4
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0987s; samplesPerSecond = 6481.8
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0986s; samplesPerSecond = 6490.3
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.1599s
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.0983s; samplesPerSecond = 6509.8
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0985s; samplesPerSecond = 6496.4
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0985s; samplesPerSecond = 6495.7
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0986s; samplesPerSecond = 6487.8
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0986s; samplesPerSecond = 6488.5
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0987s; samplesPerSecond = 6485.1
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0987s; samplesPerSecond = 6487.4
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.0986s; samplesPerSecond = 6490.1
|
||||
MPI Rank 1: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0986s; samplesPerSecond = 6491.3
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0987s; samplesPerSecond = 6486.6
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0987s; samplesPerSecond = 6485.0
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0986s; samplesPerSecond = 6493.4
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0985s; samplesPerSecond = 6495.7
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0986s; samplesPerSecond = 6491.1
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6487.2
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0988s; samplesPerSecond = 6480.5
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0987s; samplesPerSecond = 6487.1
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0987s; samplesPerSecond = 6487.6
|
||||
MPI Rank 1: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0987s; samplesPerSecond = 6485.7
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0987s; samplesPerSecond = 6487.3
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0987s; samplesPerSecond = 6483.9
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6491.6
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6488.6
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0987s; samplesPerSecond = 6485.3
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0987s; samplesPerSecond = 6482.3
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0988s; samplesPerSecond = 6481.0
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0987s; samplesPerSecond = 6485.5
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0987s; samplesPerSecond = 6484.0
|
||||
MPI Rank 1: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0987s; samplesPerSecond = 6482.3
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0987s; samplesPerSecond = 6484.4
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0987s; samplesPerSecond = 6481.8
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0986s; samplesPerSecond = 6490.3
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.1599s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1246,45 +1246,45 @@ MPI Rank 1: 08/16/2016 10:09:37: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.017293
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.001855
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.011879
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalErrorPrediction = 0.61111111 * 2304; time = 0.1389s; samplesPerSecond = 16591.8
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalClassificationError = 0.61111111 * 2304; time = 0.1389s; samplesPerSecond = 16591.8
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.00617
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.014926
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003283
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.014854
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.1366s; samplesPerSecond = 18737.1
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.1366s; samplesPerSecond = 18737.1
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.002261
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.011359
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.002224
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.011379
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalErrorPrediction = 0.57617188 * 2560; time = 0.1327s; samplesPerSecond = 19297.9
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalClassificationError = 0.57617188 * 2560; time = 0.1327s; samplesPerSecond = 19297.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.001214
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.012939
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.002258
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.011643
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalErrorPrediction = 0.60664063 * 2560; time = 0.1332s; samplesPerSecond = 19214.7
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalClassificationError = 0.60664063 * 2560; time = 0.1332s; samplesPerSecond = 19214.7
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.002313
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.011347
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.00226
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.011452
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalErrorPrediction = 0.58945313 * 2560; time = 0.1328s; samplesPerSecond = 19271.6
|
||||
MPI Rank 1: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalClassificationError = 0.58945313 * 2560; time = 0.1328s; samplesPerSecond = 19271.6
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.002246
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.011362
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003668
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.013309
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalErrorPrediction = 0.56562500 * 2560; time = 0.1344s; samplesPerSecond = 19041.0
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalClassificationError = 0.56562500 * 2560; time = 0.1344s; samplesPerSecond = 19041.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.001666
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.01288
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.001214
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.012966
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalErrorPrediction = 0.59179688 * 2560; time = 0.1327s; samplesPerSecond = 19288.3
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalClassificationError = 0.59179688 * 2560; time = 0.1327s; samplesPerSecond = 19288.3
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.001229
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.013198
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.005102
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.011467
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalErrorPrediction = 0.58632812 * 2560; time = 0.1331s; samplesPerSecond = 19235.7
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalClassificationError = 0.58632812 * 2560; time = 0.1331s; samplesPerSecond = 19235.7
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003345
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.003182
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalErrorPrediction = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.08367s
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalClassificationError = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.08367s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1294,13 +1294,13 @@ MPI Rank 1: Async gradient aggregation wait time: 0.007435
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.028784
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.006185
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.027921
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalErrorPrediction = 0.56510417 * 9216; time = 0.2773s; samplesPerSecond = 33229.9
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalClassificationError = 0.56510417 * 9216; time = 0.2773s; samplesPerSecond = 33229.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.001133
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.027435
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.000407
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.022388
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalErrorPrediction = 0.56914062 * 10240; time = 0.2770s; samplesPerSecond = 36970.2
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalErrorPrediction = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.562969s
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalClassificationError = 0.56914062 * 10240; time = 0.2770s; samplesPerSecond = 36970.2
|
||||
MPI Rank 1: 08/16/2016 10:09:38: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalClassificationError = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.562969s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:39: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1310,14 +1310,14 @@ MPI Rank 1: Async gradient aggregation wait time: 0.001906
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.027016
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003939
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.025318
|
||||
MPI Rank 1: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalErrorPrediction = 0.53982205 * 9216; time = 0.2719s; samplesPerSecond = 33897.4
|
||||
MPI Rank 1: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalClassificationError = 0.53982205 * 9216; time = 0.2719s; samplesPerSecond = 33897.4
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.001777
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.02671
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.001862
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.026708
|
||||
MPI Rank 1: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalErrorPrediction = 0.53369141 * 10240; time = 0.2738s; samplesPerSecond = 37402.6
|
||||
MPI Rank 1: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalClassificationError = 0.53369141 * 10240; time = 0.2738s; samplesPerSecond = 37402.6
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003371
|
||||
MPI Rank 1: 08/16/2016 10:09:39: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalErrorPrediction = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.555452s
|
||||
MPI Rank 1: 08/16/2016 10:09:39: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalClassificationError = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.555452s
|
||||
MPI Rank 1: 08/16/2016 10:09:39: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:09:39: Action "train" complete.
|
||||
|
|
@ -1369,7 +1369,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1395,7 +1395,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1478,7 +1478,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1504,7 +1504,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1593,7 +1593,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1619,7 +1619,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1716,7 +1716,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1745,7 +1745,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1768,7 +1768,7 @@ MPI Rank 2: 08/16/2016 10:09:31: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 10:09:31: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:31: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 10:09:31: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 10:09:31: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1821,39 +1821,39 @@ MPI Rank 2: 08/16/2016 10:09:34: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.0993s; samplesPerSecond = 6442.8
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0985s; samplesPerSecond = 6495.7
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0985s; samplesPerSecond = 6496.7
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0986s; samplesPerSecond = 6488.2
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0986s; samplesPerSecond = 6489.8
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0987s; samplesPerSecond = 6485.2
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0987s; samplesPerSecond = 6487.2
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.0986s; samplesPerSecond = 6490.4
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0986s; samplesPerSecond = 6491.7
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0987s; samplesPerSecond = 6487.3
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0987s; samplesPerSecond = 6485.3
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0985s; samplesPerSecond = 6494.7
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0985s; samplesPerSecond = 6496.6
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0986s; samplesPerSecond = 6491.3
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6487.3
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0987s; samplesPerSecond = 6481.0
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0987s; samplesPerSecond = 6487.2
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0987s; samplesPerSecond = 6487.6
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0987s; samplesPerSecond = 6485.8
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0987s; samplesPerSecond = 6487.3
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0987s; samplesPerSecond = 6484.8
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6491.1
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6489.4
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0987s; samplesPerSecond = 6485.2
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0987s; samplesPerSecond = 6482.8
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0987s; samplesPerSecond = 6482.3
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0987s; samplesPerSecond = 6485.6
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0987s; samplesPerSecond = 6484.2
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0987s; samplesPerSecond = 6483.4
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0987s; samplesPerSecond = 6483.9
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0987s; samplesPerSecond = 6481.9
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0986s; samplesPerSecond = 6490.0
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16101s
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.0993s; samplesPerSecond = 6442.8
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0985s; samplesPerSecond = 6495.7
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0985s; samplesPerSecond = 6496.7
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0986s; samplesPerSecond = 6488.2
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0986s; samplesPerSecond = 6489.8
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0987s; samplesPerSecond = 6485.2
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0987s; samplesPerSecond = 6487.2
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.0986s; samplesPerSecond = 6490.4
|
||||
MPI Rank 2: 08/16/2016 10:09:34: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0986s; samplesPerSecond = 6491.7
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0987s; samplesPerSecond = 6487.3
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0987s; samplesPerSecond = 6485.3
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0985s; samplesPerSecond = 6494.7
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0985s; samplesPerSecond = 6496.6
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0986s; samplesPerSecond = 6491.3
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6487.3
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0987s; samplesPerSecond = 6481.0
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0987s; samplesPerSecond = 6487.2
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0987s; samplesPerSecond = 6487.6
|
||||
MPI Rank 2: 08/16/2016 10:09:35: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0987s; samplesPerSecond = 6485.8
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0987s; samplesPerSecond = 6487.3
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0987s; samplesPerSecond = 6484.8
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6491.1
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0986s; samplesPerSecond = 6489.4
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0987s; samplesPerSecond = 6485.2
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0987s; samplesPerSecond = 6482.8
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0987s; samplesPerSecond = 6482.3
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0987s; samplesPerSecond = 6485.6
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0987s; samplesPerSecond = 6484.2
|
||||
MPI Rank 2: 08/16/2016 10:09:36: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0987s; samplesPerSecond = 6483.4
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0987s; samplesPerSecond = 6483.9
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0987s; samplesPerSecond = 6481.9
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0986s; samplesPerSecond = 6490.0
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16101s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1862,45 +1862,45 @@ MPI Rank 2: 08/16/2016 10:09:37: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.014665
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.001294
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011743
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalErrorPrediction = 0.61111111 * 2304; time = 0.1486s; samplesPerSecond = 15506.6
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalClassificationError = 0.61111111 * 2304; time = 0.1486s; samplesPerSecond = 15506.6
|
||||
MPI Rank 2: Async gradient aggregation wait time: 2e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.00478
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.005358
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.014693
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.1270s; samplesPerSecond = 20156.2
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.1270s; samplesPerSecond = 20156.2
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.001694
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011815
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.002553
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011853
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalErrorPrediction = 0.57617188 * 2560; time = 0.1322s; samplesPerSecond = 19360.8
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalClassificationError = 0.57617188 * 2560; time = 0.1322s; samplesPerSecond = 19360.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.003978
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011672
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.0017
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011758
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalErrorPrediction = 0.60664063 * 2560; time = 0.1336s; samplesPerSecond = 19164.7
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalClassificationError = 0.60664063 * 2560; time = 0.1336s; samplesPerSecond = 19164.7
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.002587
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011795
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.001696
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011901
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalErrorPrediction = 0.58945313 * 2560; time = 0.1329s; samplesPerSecond = 19257.7
|
||||
MPI Rank 2: 08/16/2016 10:09:37: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalClassificationError = 0.58945313 * 2560; time = 0.1329s; samplesPerSecond = 19257.7
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.002555
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011821
|
||||
MPI Rank 2: Async gradient aggregation wait time: 2e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.005946
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalErrorPrediction = 0.56562500 * 2560; time = 0.1339s; samplesPerSecond = 19111.6
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalClassificationError = 0.56562500 * 2560; time = 0.1339s; samplesPerSecond = 19111.6
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.001929
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011609
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.003976
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.01164
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalErrorPrediction = 0.59179688 * 2560; time = 0.1328s; samplesPerSecond = 19270.3
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalClassificationError = 0.59179688 * 2560; time = 0.1328s; samplesPerSecond = 19270.3
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.00398
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011589
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.000283
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011927
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalErrorPrediction = 0.58632812 * 2560; time = 0.1332s; samplesPerSecond = 19224.6
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalClassificationError = 0.58632812 * 2560; time = 0.1332s; samplesPerSecond = 19224.6
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.003336
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.004097
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalErrorPrediction = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.08382s
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalClassificationError = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.08382s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1910,13 +1910,13 @@ MPI Rank 2: Async gradient aggregation wait time: 3e-06
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.022531
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.011564
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.026393
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalErrorPrediction = 0.56510417 * 9216; time = 0.2777s; samplesPerSecond = 33187.8
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalClassificationError = 0.56510417 * 9216; time = 0.2777s; samplesPerSecond = 33187.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.006793
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.026043
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.007443
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.0209
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalErrorPrediction = 0.56914062 * 10240; time = 0.2699s; samplesPerSecond = 37934.9
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalErrorPrediction = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.563163s
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalClassificationError = 0.56914062 * 10240; time = 0.2699s; samplesPerSecond = 37934.9
|
||||
MPI Rank 2: 08/16/2016 10:09:38: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalClassificationError = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.563163s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:39: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1926,14 +1926,14 @@ MPI Rank 2: Async gradient aggregation wait time: 0.015928
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.027468
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.001119
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.025775
|
||||
MPI Rank 2: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalErrorPrediction = 0.53982205 * 9216; time = 0.2717s; samplesPerSecond = 33924.8
|
||||
MPI Rank 2: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalClassificationError = 0.53982205 * 9216; time = 0.2717s; samplesPerSecond = 33924.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.006372
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.027123
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004274
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.027133
|
||||
MPI Rank 2: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalErrorPrediction = 0.53369141 * 10240; time = 0.2739s; samplesPerSecond = 37391.9
|
||||
MPI Rank 2: 08/16/2016 10:09:39: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalClassificationError = 0.53369141 * 10240; time = 0.2739s; samplesPerSecond = 37391.9
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.003741
|
||||
MPI Rank 2: 08/16/2016 10:09:39: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalErrorPrediction = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.555698s
|
||||
MPI Rank 2: 08/16/2016 10:09:39: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalClassificationError = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.555698s
|
||||
MPI Rank 2: 08/16/2016 10:09:39: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:09:39: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -131,7 +131,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -157,7 +157,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -240,7 +240,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -266,7 +266,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -355,7 +355,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -381,7 +381,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -477,7 +477,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -506,7 +506,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -529,7 +529,7 @@ MPI Rank 0: 08/16/2016 03:20:21: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:20:21: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:21: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:20:21: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:20:21: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -582,39 +582,39 @@ MPI Rank 0: 08/16/2016 03:20:25: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:25: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.1303s; samplesPerSecond = 4912.5
|
||||
MPI Rank 0: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.0982s; samplesPerSecond = 6519.1
|
||||
MPI Rank 0: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.1010s; samplesPerSecond = 6337.4
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0989s; samplesPerSecond = 6472.8
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0989s; samplesPerSecond = 6474.3
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.0983s; samplesPerSecond = 6508.8
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.0981s; samplesPerSecond = 6523.2
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.1019s; samplesPerSecond = 6278.6
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.0986s; samplesPerSecond = 6491.3
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.0946s; samplesPerSecond = 6764.2
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.0991s; samplesPerSecond = 6458.1
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.1071s; samplesPerSecond = 5976.8
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.0966s; samplesPerSecond = 6627.8
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.1014s; samplesPerSecond = 6308.6
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.1014s; samplesPerSecond = 6312.1
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.0989s; samplesPerSecond = 6470.5
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.0942s; samplesPerSecond = 6793.3
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.0959s; samplesPerSecond = 6674.6
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.1003s; samplesPerSecond = 6382.8
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.0984s; samplesPerSecond = 6503.8
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.1006s; samplesPerSecond = 6364.0
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.1037s; samplesPerSecond = 6172.0
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.1099s; samplesPerSecond = 5821.0
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1020s; samplesPerSecond = 6273.3
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.0998s; samplesPerSecond = 6412.6
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.0967s; samplesPerSecond = 6616.2
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0990s; samplesPerSecond = 6463.7
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1046s; samplesPerSecond = 6118.4
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1031s; samplesPerSecond = 6208.1
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.1083s; samplesPerSecond = 5908.9
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.1037s; samplesPerSecond = 6169.2
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.0942s; samplesPerSecond = 6793.7
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.26543s
|
||||
MPI Rank 0: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.1303s; samplesPerSecond = 4912.5
|
||||
MPI Rank 0: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.0982s; samplesPerSecond = 6519.1
|
||||
MPI Rank 0: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.1010s; samplesPerSecond = 6337.4
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0989s; samplesPerSecond = 6472.8
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0989s; samplesPerSecond = 6474.3
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.0983s; samplesPerSecond = 6508.8
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.0981s; samplesPerSecond = 6523.2
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.1019s; samplesPerSecond = 6278.6
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.0986s; samplesPerSecond = 6491.3
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.0946s; samplesPerSecond = 6764.2
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.0991s; samplesPerSecond = 6458.1
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.1071s; samplesPerSecond = 5976.8
|
||||
MPI Rank 0: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.0966s; samplesPerSecond = 6627.8
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.1014s; samplesPerSecond = 6308.6
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.1014s; samplesPerSecond = 6312.1
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.0989s; samplesPerSecond = 6470.5
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.0942s; samplesPerSecond = 6793.3
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.0959s; samplesPerSecond = 6674.6
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.1003s; samplesPerSecond = 6382.8
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.0984s; samplesPerSecond = 6503.8
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.1006s; samplesPerSecond = 6364.0
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.1037s; samplesPerSecond = 6172.0
|
||||
MPI Rank 0: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.1099s; samplesPerSecond = 5821.0
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1020s; samplesPerSecond = 6273.3
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.0998s; samplesPerSecond = 6412.6
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.0967s; samplesPerSecond = 6616.2
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0990s; samplesPerSecond = 6463.7
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1046s; samplesPerSecond = 6118.4
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1031s; samplesPerSecond = 6208.1
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.1083s; samplesPerSecond = 5908.9
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.1037s; samplesPerSecond = 6169.2
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.0942s; samplesPerSecond = 6793.7
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.26543s
|
||||
MPI Rank 0: 08/16/2016 03:20:28: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:28: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -624,45 +624,45 @@ MPI Rank 0: 08/16/2016 03:20:28: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.017461
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.004531
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021009
|
||||
MPI Rank 0: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17852518 * 2304; EvalErrorPrediction = 0.60720486 * 2304; time = 0.2252s; samplesPerSecond = 10232.5
|
||||
MPI Rank 0: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17852518 * 2304; EvalClassificationError = 0.60720486 * 2304; time = 0.2252s; samplesPerSecond = 10232.5
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.005597
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021356
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002771
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.02228
|
||||
MPI Rank 0: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.20195382 * 2560; EvalErrorPrediction = 0.57968750 * 2560; time = 0.2205s; samplesPerSecond = 11609.8
|
||||
MPI Rank 0: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.20195382 * 2560; EvalClassificationError = 0.57968750 * 2560; time = 0.2205s; samplesPerSecond = 11609.8
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.006954
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021547
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.007574
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021451
|
||||
MPI Rank 0: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.13354572 * 2560; EvalErrorPrediction = 0.57226563 * 2560; time = 0.2176s; samplesPerSecond = 11764.2
|
||||
MPI Rank 0: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.13354572 * 2560; EvalClassificationError = 0.57226563 * 2560; time = 0.2176s; samplesPerSecond = 11764.2
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.006622
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021298
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.007007
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021108
|
||||
MPI Rank 0: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.13570097 * 2560; EvalErrorPrediction = 0.59804687 * 2560; time = 0.2126s; samplesPerSecond = 12040.2
|
||||
MPI Rank 0: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.13570097 * 2560; EvalClassificationError = 0.59804687 * 2560; time = 0.2126s; samplesPerSecond = 12040.2
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00622
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.02044
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.007231
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021619
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17201001 * 2560; EvalErrorPrediction = 0.59453125 * 2560; time = 0.2155s; samplesPerSecond = 11878.4
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17201001 * 2560; EvalClassificationError = 0.59453125 * 2560; time = 0.2155s; samplesPerSecond = 11878.4
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.006772
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.022132
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.009431
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.020768
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04702260 * 2560; EvalErrorPrediction = 0.56406250 * 2560; time = 0.2164s; samplesPerSecond = 11830.8
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04702260 * 2560; EvalClassificationError = 0.56406250 * 2560; time = 0.2164s; samplesPerSecond = 11830.8
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.006197
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021308
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.007342
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.020112
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11116165 * 2560; EvalErrorPrediction = 0.62031250 * 2560; time = 0.2135s; samplesPerSecond = 11991.0
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11116165 * 2560; EvalClassificationError = 0.62031250 * 2560; time = 0.2135s; samplesPerSecond = 11991.0
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.005218
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.020588
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.008348
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021196
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06023983 * 2560; EvalErrorPrediction = 0.58437500 * 2560; time = 0.2177s; samplesPerSecond = 11761.2
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06023983 * 2560; EvalClassificationError = 0.58437500 * 2560; time = 0.2177s; samplesPerSecond = 11761.2
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.017768
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021528
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.13051935 * 20480; EvalErrorPrediction = 0.58984375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.78762s
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.13051935 * 20480; EvalClassificationError = 0.58984375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.78762s
|
||||
MPI Rank 0: 08/16/2016 03:20:30: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:30: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -673,13 +673,13 @@ MPI Rank 0: Async gradient aggregation wait time: 3e-006
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.020512
|
||||
MPI Rank 0: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.021598
|
||||
MPI Rank 0: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.18543157 * 9216; EvalErrorPrediction = 0.57552083 * 9216; time = 0.5199s; samplesPerSecond = 17725.8
|
||||
MPI Rank 0: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.18543157 * 9216; EvalClassificationError = 0.57552083 * 9216; time = 0.5199s; samplesPerSecond = 17725.8
|
||||
MPI Rank 0: Async gradient aggregation wait time: 4e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.028821
|
||||
MPI Rank 0: Async gradient aggregation wait time: 4e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.024884
|
||||
MPI Rank 0: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.09876585 * 10240; EvalErrorPrediction = 0.57324219 * 10240; time = 0.4745s; samplesPerSecond = 21579.0
|
||||
MPI Rank 0: 08/16/2016 03:20:31: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.13466219 * 20480; EvalErrorPrediction = 0.57324219 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.0605s
|
||||
MPI Rank 0: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.09876585 * 10240; EvalClassificationError = 0.57324219 * 10240; time = 0.4745s; samplesPerSecond = 21579.0
|
||||
MPI Rank 0: 08/16/2016 03:20:31: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.13466219 * 20480; EvalClassificationError = 0.57324219 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.0605s
|
||||
MPI Rank 0: 08/16/2016 03:20:32: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:32: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -690,14 +690,14 @@ MPI Rank 0: Async gradient aggregation wait time: 5e-006
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.018185
|
||||
MPI Rank 0: Async gradient aggregation wait time: 4e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.020773
|
||||
MPI Rank 0: 08/16/2016 03:20:32: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.96095323 * 9216; EvalErrorPrediction = 0.53678385 * 9216; time = 0.4751s; samplesPerSecond = 19396.8
|
||||
MPI Rank 0: 08/16/2016 03:20:32: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.96095323 * 9216; EvalClassificationError = 0.53678385 * 9216; time = 0.4751s; samplesPerSecond = 19396.8
|
||||
MPI Rank 0: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.020298
|
||||
MPI Rank 0: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.049557
|
||||
MPI Rank 0: 08/16/2016 03:20:33: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93674809 * 10240; EvalErrorPrediction = 0.52773437 * 10240; time = 0.4668s; samplesPerSecond = 21938.6
|
||||
MPI Rank 0: 08/16/2016 03:20:33: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93674809 * 10240; EvalClassificationError = 0.52773437 * 10240; time = 0.4668s; samplesPerSecond = 21938.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.021073
|
||||
MPI Rank 0: 08/16/2016 03:20:33: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94894630 * 20480; EvalErrorPrediction = 0.53217773 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=1.01124s
|
||||
MPI Rank 0: 08/16/2016 03:20:33: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94894630 * 20480; EvalClassificationError = 0.53217773 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=1.01124s
|
||||
MPI Rank 0: 08/16/2016 03:20:33: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:20:33: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -751,7 +751,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -777,7 +777,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -860,7 +860,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -886,7 +886,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -975,7 +975,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1001,7 +1001,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1097,7 +1097,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1126,7 +1126,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1149,7 +1149,7 @@ MPI Rank 1: 08/16/2016 03:20:22: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:20:22: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:22: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:20:22: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:20:22: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1202,39 +1202,39 @@ MPI Rank 1: 08/16/2016 03:20:25: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:25: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.1299s; samplesPerSecond = 4926.1
|
||||
MPI Rank 1: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.1031s; samplesPerSecond = 6209.3
|
||||
MPI Rank 1: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.0965s; samplesPerSecond = 6629.1
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1000s; samplesPerSecond = 6401.5
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0984s; samplesPerSecond = 6502.3
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.0997s; samplesPerSecond = 6417.8
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.1021s; samplesPerSecond = 6268.1
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.1023s; samplesPerSecond = 6255.7
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.1069s; samplesPerSecond = 5987.6
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.1015s; samplesPerSecond = 6306.5
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.1002s; samplesPerSecond = 6389.0
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0997s; samplesPerSecond = 6422.3
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.1008s; samplesPerSecond = 6346.6
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.1013s; samplesPerSecond = 6318.4
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.1029s; samplesPerSecond = 6221.4
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.1017s; samplesPerSecond = 6289.9
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.1053s; samplesPerSecond = 6076.5
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.1008s; samplesPerSecond = 6347.6
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.1016s; samplesPerSecond = 6301.3
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.1010s; samplesPerSecond = 6336.7
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.1001s; samplesPerSecond = 6395.9
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.1013s; samplesPerSecond = 6317.4
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0996s; samplesPerSecond = 6427.0
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1026s; samplesPerSecond = 6236.2
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.1030s; samplesPerSecond = 6213.9
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.0996s; samplesPerSecond = 6426.2
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.1010s; samplesPerSecond = 6337.4
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0968s; samplesPerSecond = 6609.0
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1075s; samplesPerSecond = 5952.2
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.1043s; samplesPerSecond = 6134.7
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0993s; samplesPerSecond = 6443.4
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.0936s; samplesPerSecond = 6841.0
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.2916s
|
||||
MPI Rank 1: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.1299s; samplesPerSecond = 4926.1
|
||||
MPI Rank 1: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.1031s; samplesPerSecond = 6209.3
|
||||
MPI Rank 1: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.0965s; samplesPerSecond = 6629.1
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1000s; samplesPerSecond = 6401.5
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0984s; samplesPerSecond = 6502.3
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.0997s; samplesPerSecond = 6417.8
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.1021s; samplesPerSecond = 6268.1
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.1023s; samplesPerSecond = 6255.7
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.1069s; samplesPerSecond = 5987.6
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.1015s; samplesPerSecond = 6306.5
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.1002s; samplesPerSecond = 6389.0
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0997s; samplesPerSecond = 6422.3
|
||||
MPI Rank 1: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.1008s; samplesPerSecond = 6346.6
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.1013s; samplesPerSecond = 6318.4
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.1029s; samplesPerSecond = 6221.4
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.1017s; samplesPerSecond = 6289.9
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.1053s; samplesPerSecond = 6076.5
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.1008s; samplesPerSecond = 6347.6
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.1016s; samplesPerSecond = 6301.3
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.1010s; samplesPerSecond = 6336.7
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.1001s; samplesPerSecond = 6395.9
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.1013s; samplesPerSecond = 6317.4
|
||||
MPI Rank 1: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0996s; samplesPerSecond = 6427.0
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1026s; samplesPerSecond = 6236.2
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.1030s; samplesPerSecond = 6213.9
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.0996s; samplesPerSecond = 6426.2
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.1010s; samplesPerSecond = 6337.4
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0968s; samplesPerSecond = 6609.0
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1075s; samplesPerSecond = 5952.2
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.1043s; samplesPerSecond = 6134.7
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0993s; samplesPerSecond = 6443.4
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.0936s; samplesPerSecond = 6841.0
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.2916s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:28: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1243,45 +1243,45 @@ MPI Rank 1: 08/16/2016 03:20:28: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.029656
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007273
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021183
|
||||
MPI Rank 1: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17852518 * 2304; EvalErrorPrediction = 0.60720486 * 2304; time = 0.2284s; samplesPerSecond = 10087.0
|
||||
MPI Rank 1: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17852518 * 2304; EvalClassificationError = 0.60720486 * 2304; time = 0.2284s; samplesPerSecond = 10087.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007317
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.023177
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007913
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021791
|
||||
MPI Rank 1: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.20195382 * 2560; EvalErrorPrediction = 0.57968750 * 2560; time = 0.2213s; samplesPerSecond = 11567.5
|
||||
MPI Rank 1: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.20195382 * 2560; EvalClassificationError = 0.57968750 * 2560; time = 0.2213s; samplesPerSecond = 11567.5
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.006082
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021886
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.006563
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021874
|
||||
MPI Rank 1: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.13354572 * 2560; EvalErrorPrediction = 0.57226563 * 2560; time = 0.2158s; samplesPerSecond = 11860.5
|
||||
MPI Rank 1: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.13354572 * 2560; EvalClassificationError = 0.57226563 * 2560; time = 0.2158s; samplesPerSecond = 11860.5
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007618
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021328
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007356
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.020426
|
||||
MPI Rank 1: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.13570097 * 2560; EvalErrorPrediction = 0.59804687 * 2560; time = 0.2127s; samplesPerSecond = 12033.0
|
||||
MPI Rank 1: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.13570097 * 2560; EvalClassificationError = 0.59804687 * 2560; time = 0.2127s; samplesPerSecond = 12033.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007275
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.020417
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.008568
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.022891
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17201001 * 2560; EvalErrorPrediction = 0.59453125 * 2560; time = 0.2161s; samplesPerSecond = 11848.9
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17201001 * 2560; EvalClassificationError = 0.59453125 * 2560; time = 0.2161s; samplesPerSecond = 11848.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.00506
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021379
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.011516
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.022444
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04702260 * 2560; EvalErrorPrediction = 0.56406250 * 2560; time = 0.2152s; samplesPerSecond = 11897.0
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04702260 * 2560; EvalClassificationError = 0.56406250 * 2560; time = 0.2152s; samplesPerSecond = 11897.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.005813
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.022097
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.008247
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021675
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11116165 * 2560; EvalErrorPrediction = 0.62031250 * 2560; time = 0.2141s; samplesPerSecond = 11958.1
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11116165 * 2560; EvalClassificationError = 0.62031250 * 2560; time = 0.2141s; samplesPerSecond = 11958.1
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.009639
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.022288
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007823
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021267
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06023983 * 2560; EvalErrorPrediction = 0.58437500 * 2560; time = 0.2160s; samplesPerSecond = 11849.2
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06023983 * 2560; EvalClassificationError = 0.58437500 * 2560; time = 0.2160s; samplesPerSecond = 11849.2
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.019456
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.022119
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.13051935 * 20480; EvalErrorPrediction = 0.58984375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.79029s
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.13051935 * 20480; EvalClassificationError = 0.58984375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.79029s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:30: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1291,13 +1291,13 @@ MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.039428
|
||||
MPI Rank 1: Async gradient aggregation wait time: 7e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.035112
|
||||
MPI Rank 1: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.18543157 * 9216; EvalErrorPrediction = 0.57552083 * 9216; time = 0.5127s; samplesPerSecond = 17974.0
|
||||
MPI Rank 1: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.18543157 * 9216; EvalClassificationError = 0.57552083 * 9216; time = 0.5127s; samplesPerSecond = 17974.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 4e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021272
|
||||
MPI Rank 1: Async gradient aggregation wait time: 4e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.019826
|
||||
MPI Rank 1: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.09876585 * 10240; EvalErrorPrediction = 0.57324219 * 10240; time = 0.5042s; samplesPerSecond = 20308.5
|
||||
MPI Rank 1: 08/16/2016 03:20:31: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.13466219 * 20480; EvalErrorPrediction = 0.57324219 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.06315s
|
||||
MPI Rank 1: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.09876585 * 10240; EvalClassificationError = 0.57324219 * 10240; time = 0.5042s; samplesPerSecond = 20308.5
|
||||
MPI Rank 1: 08/16/2016 03:20:31: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.13466219 * 20480; EvalClassificationError = 0.57324219 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.06315s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:32: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1307,14 +1307,14 @@ MPI Rank 1: Async gradient aggregation wait time: 4e-006
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.032424
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.002787
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.042553
|
||||
MPI Rank 1: 08/16/2016 03:20:32: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.96095323 * 9216; EvalErrorPrediction = 0.53678385 * 9216; time = 0.4556s; samplesPerSecond = 20229.3
|
||||
MPI Rank 1: 08/16/2016 03:20:32: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.96095323 * 9216; EvalClassificationError = 0.53678385 * 9216; time = 0.4556s; samplesPerSecond = 20229.3
|
||||
MPI Rank 1: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.028505
|
||||
MPI Rank 1: Async gradient aggregation wait time: 4e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021643
|
||||
MPI Rank 1: 08/16/2016 03:20:33: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93674809 * 10240; EvalErrorPrediction = 0.52773437 * 10240; time = 0.5099s; samplesPerSecond = 20081.2
|
||||
MPI Rank 1: 08/16/2016 03:20:33: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93674809 * 10240; EvalClassificationError = 0.52773437 * 10240; time = 0.5099s; samplesPerSecond = 20081.2
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.02112
|
||||
MPI Rank 1: 08/16/2016 03:20:33: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94894630 * 20480; EvalErrorPrediction = 0.53217773 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=1.01391s
|
||||
MPI Rank 1: 08/16/2016 03:20:33: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94894630 * 20480; EvalClassificationError = 0.53217773 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=1.01391s
|
||||
MPI Rank 1: 08/16/2016 03:20:33: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:33: Action "train" complete.
|
||||
|
|
@ -1367,7 +1367,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1393,7 +1393,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1476,7 +1476,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1502,7 +1502,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1591,7 +1591,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1617,7 +1617,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1713,7 +1713,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1742,7 +1742,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1765,7 +1765,7 @@ MPI Rank 2: 08/16/2016 03:20:22: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 03:20:22: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:22: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 03:20:22: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 03:20:22: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1818,39 +1818,39 @@ MPI Rank 2: 08/16/2016 03:20:25: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:25: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.1302s; samplesPerSecond = 4916.5
|
||||
MPI Rank 2: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.0975s; samplesPerSecond = 6567.1
|
||||
MPI Rank 2: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.1016s; samplesPerSecond = 6298.2
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0994s; samplesPerSecond = 6436.7
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.1035s; samplesPerSecond = 6181.3
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.1044s; samplesPerSecond = 6132.0
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.1029s; samplesPerSecond = 6220.3
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.1010s; samplesPerSecond = 6334.4
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.0996s; samplesPerSecond = 6424.0
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.0978s; samplesPerSecond = 6545.3
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.0994s; samplesPerSecond = 6441.4
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0983s; samplesPerSecond = 6512.1
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.0963s; samplesPerSecond = 6644.1
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.0993s; samplesPerSecond = 6443.1
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.0987s; samplesPerSecond = 6481.3
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.0973s; samplesPerSecond = 6577.8
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.0966s; samplesPerSecond = 6626.6
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.0955s; samplesPerSecond = 6704.4
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.0942s; samplesPerSecond = 6797.4
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.0971s; samplesPerSecond = 6590.6
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.1019s; samplesPerSecond = 6282.2
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.1032s; samplesPerSecond = 6199.3
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0984s; samplesPerSecond = 6500.9
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1021s; samplesPerSecond = 6267.7
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.0991s; samplesPerSecond = 6458.3
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.1003s; samplesPerSecond = 6379.8
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0983s; samplesPerSecond = 6513.2
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0948s; samplesPerSecond = 6748.3
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0947s; samplesPerSecond = 6761.7
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0928s; samplesPerSecond = 6898.5
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0913s; samplesPerSecond = 7009.0
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.0916s; samplesPerSecond = 6990.2
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.20727s
|
||||
MPI Rank 2: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.1302s; samplesPerSecond = 4916.5
|
||||
MPI Rank 2: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.0975s; samplesPerSecond = 6567.1
|
||||
MPI Rank 2: 08/16/2016 03:20:25: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.1016s; samplesPerSecond = 6298.2
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0994s; samplesPerSecond = 6436.7
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.1035s; samplesPerSecond = 6181.3
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.1044s; samplesPerSecond = 6132.0
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.1029s; samplesPerSecond = 6220.3
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.1010s; samplesPerSecond = 6334.4
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.0996s; samplesPerSecond = 6424.0
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.0978s; samplesPerSecond = 6545.3
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.0994s; samplesPerSecond = 6441.4
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0983s; samplesPerSecond = 6512.1
|
||||
MPI Rank 2: 08/16/2016 03:20:26: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.0963s; samplesPerSecond = 6644.1
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.0993s; samplesPerSecond = 6443.1
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.0987s; samplesPerSecond = 6481.3
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.0973s; samplesPerSecond = 6577.8
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.0966s; samplesPerSecond = 6626.6
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.0955s; samplesPerSecond = 6704.4
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.0942s; samplesPerSecond = 6797.4
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.0971s; samplesPerSecond = 6590.6
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.1019s; samplesPerSecond = 6282.2
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.1032s; samplesPerSecond = 6199.3
|
||||
MPI Rank 2: 08/16/2016 03:20:27: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0984s; samplesPerSecond = 6500.9
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1021s; samplesPerSecond = 6267.7
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.0991s; samplesPerSecond = 6458.3
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.1003s; samplesPerSecond = 6379.8
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0983s; samplesPerSecond = 6513.2
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0948s; samplesPerSecond = 6748.3
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0947s; samplesPerSecond = 6761.7
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0928s; samplesPerSecond = 6898.5
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0913s; samplesPerSecond = 7009.0
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.0916s; samplesPerSecond = 6990.2
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.20727s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:28: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1859,45 +1859,45 @@ MPI Rank 2: 08/16/2016 03:20:28: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.032204
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.010081
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.021164
|
||||
MPI Rank 2: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17852518 * 2304; EvalErrorPrediction = 0.60720486 * 2304; time = 0.2294s; samplesPerSecond = 10045.7
|
||||
MPI Rank 2: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17852518 * 2304; EvalClassificationError = 0.60720486 * 2304; time = 0.2294s; samplesPerSecond = 10045.7
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.006745
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.023171
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.013627
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.020912
|
||||
MPI Rank 2: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.20195382 * 2560; EvalErrorPrediction = 0.57968750 * 2560; time = 0.2212s; samplesPerSecond = 11573.6
|
||||
MPI Rank 2: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.20195382 * 2560; EvalClassificationError = 0.57968750 * 2560; time = 0.2212s; samplesPerSecond = 11573.6
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.009276
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.021067
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.010184
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.02104
|
||||
MPI Rank 2: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.13354572 * 2560; EvalErrorPrediction = 0.57226563 * 2560; time = 0.2154s; samplesPerSecond = 11884.9
|
||||
MPI Rank 2: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.13354572 * 2560; EvalClassificationError = 0.57226563 * 2560; time = 0.2154s; samplesPerSecond = 11884.9
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.009262
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.021326
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.006774
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.021214
|
||||
MPI Rank 2: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.13570097 * 2560; EvalErrorPrediction = 0.59804687 * 2560; time = 0.2124s; samplesPerSecond = 12052.2
|
||||
MPI Rank 2: 08/16/2016 03:20:29: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.13570097 * 2560; EvalClassificationError = 0.59804687 * 2560; time = 0.2124s; samplesPerSecond = 12052.2
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.007039
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.020401
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.009992
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.022942
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17201001 * 2560; EvalErrorPrediction = 0.59453125 * 2560; time = 0.2168s; samplesPerSecond = 11808.9
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17201001 * 2560; EvalClassificationError = 0.59453125 * 2560; time = 0.2168s; samplesPerSecond = 11808.9
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.01098
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.022232
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.007239
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.024066
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04702260 * 2560; EvalErrorPrediction = 0.56406250 * 2560; time = 0.2153s; samplesPerSecond = 11890.2
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.04702260 * 2560; EvalClassificationError = 0.56406250 * 2560; time = 0.2153s; samplesPerSecond = 11890.2
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.008838
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.02123
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.008702
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.024328
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11116165 * 2560; EvalErrorPrediction = 0.62031250 * 2560; time = 0.2134s; samplesPerSecond = 11993.8
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.11116165 * 2560; EvalClassificationError = 0.62031250 * 2560; time = 0.2134s; samplesPerSecond = 11993.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.00939
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.023906
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.012129
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.021258
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06023983 * 2560; EvalErrorPrediction = 0.58437500 * 2560; time = 0.2139s; samplesPerSecond = 11968.1
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.06023983 * 2560; EvalClassificationError = 0.58437500 * 2560; time = 0.2139s; samplesPerSecond = 11968.1
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.02126
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.022965
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.13051935 * 20480; EvalErrorPrediction = 0.58984375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.7907s
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.13051935 * 20480; EvalClassificationError = 0.58984375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.7907s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:30: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1907,13 +1907,13 @@ MPI Rank 2: Async gradient aggregation wait time: 0.019786
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.040852
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.024007
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.04185
|
||||
MPI Rank 2: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.18543157 * 9216; EvalErrorPrediction = 0.57552083 * 9216; time = 0.4981s; samplesPerSecond = 18500.5
|
||||
MPI Rank 2: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.18543157 * 9216; EvalClassificationError = 0.57552083 * 9216; time = 0.4981s; samplesPerSecond = 18500.5
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.009704
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.049422
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.01949
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.043551
|
||||
MPI Rank 2: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.09876585 * 10240; EvalErrorPrediction = 0.57324219 * 10240; time = 0.4827s; samplesPerSecond = 21215.3
|
||||
MPI Rank 2: 08/16/2016 03:20:31: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.13466219 * 20480; EvalErrorPrediction = 0.57324219 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.06353s
|
||||
MPI Rank 2: 08/16/2016 03:20:31: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.09876585 * 10240; EvalClassificationError = 0.57324219 * 10240; time = 0.4827s; samplesPerSecond = 21215.3
|
||||
MPI Rank 2: 08/16/2016 03:20:31: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.13466219 * 20480; EvalClassificationError = 0.57324219 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.06353s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:32: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1923,14 +1923,14 @@ MPI Rank 2: Async gradient aggregation wait time: 0.011905
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.051704
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.015128
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.042511
|
||||
MPI Rank 2: 08/16/2016 03:20:32: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.96095323 * 9216; EvalErrorPrediction = 0.53678385 * 9216; time = 0.4536s; samplesPerSecond = 20317.8
|
||||
MPI Rank 2: 08/16/2016 03:20:32: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.96095323 * 9216; EvalClassificationError = 0.53678385 * 9216; time = 0.4536s; samplesPerSecond = 20317.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.008759
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.040201
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004064
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.056603
|
||||
MPI Rank 2: 08/16/2016 03:20:33: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93674809 * 10240; EvalErrorPrediction = 0.52773437 * 10240; time = 0.4907s; samplesPerSecond = 20867.3
|
||||
MPI Rank 2: 08/16/2016 03:20:33: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93674809 * 10240; EvalClassificationError = 0.52773437 * 10240; time = 0.4907s; samplesPerSecond = 20867.3
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.02231
|
||||
MPI Rank 2: 08/16/2016 03:20:33: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94894630 * 20480; EvalErrorPrediction = 0.53217773 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=1.01433s
|
||||
MPI Rank 2: 08/16/2016 03:20:33: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94894630 * 20480; EvalClassificationError = 0.53217773 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=1.01433s
|
||||
MPI Rank 2: 08/16/2016 03:20:33: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:33: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -131,7 +131,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -157,7 +157,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -240,7 +240,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -266,7 +266,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -355,7 +355,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -381,7 +381,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -478,7 +478,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -507,7 +507,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -530,7 +530,7 @@ MPI Rank 0: 08/16/2016 03:20:44: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:20:44: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:44: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:20:44: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:20:44: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -583,39 +583,39 @@ MPI Rank 0: 08/16/2016 03:20:50: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:50: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.0848s; samplesPerSecond = 7548.6
|
||||
MPI Rank 0: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0856s; samplesPerSecond = 7479.1
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0987s; samplesPerSecond = 6482.7
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1003s; samplesPerSecond = 6382.4
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0971s; samplesPerSecond = 6589.0
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.1170s; samplesPerSecond = 5469.0
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0966s; samplesPerSecond = 6624.2
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.1007s; samplesPerSecond = 6355.6
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0970s; samplesPerSecond = 6595.0
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0989s; samplesPerSecond = 6474.5
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0895s; samplesPerSecond = 7151.2
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0982s; samplesPerSecond = 6515.7
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.1044s; samplesPerSecond = 6130.0
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0895s; samplesPerSecond = 7150.4
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.1063s; samplesPerSecond = 6020.5
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.1047s; samplesPerSecond = 6114.2
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0983s; samplesPerSecond = 6508.9
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1114s; samplesPerSecond = 5745.5
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0919s; samplesPerSecond = 6961.3
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0783s; samplesPerSecond = 8171.8
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0976s; samplesPerSecond = 6558.1
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6469.9
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1018s; samplesPerSecond = 6287.0
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0913s; samplesPerSecond = 7012.0
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.1004s; samplesPerSecond = 6375.1
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.1030s; samplesPerSecond = 6212.9
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1058s; samplesPerSecond = 6051.7
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.1072s; samplesPerSecond = 5972.8
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0980s; samplesPerSecond = 6529.9
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0954s; samplesPerSecond = 6710.5
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0957s; samplesPerSecond = 6684.5
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0673s; samplesPerSecond = 9509.2
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.12839s
|
||||
MPI Rank 0: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.0848s; samplesPerSecond = 7548.6
|
||||
MPI Rank 0: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0856s; samplesPerSecond = 7479.1
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0987s; samplesPerSecond = 6482.7
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1003s; samplesPerSecond = 6382.4
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0971s; samplesPerSecond = 6589.0
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.1170s; samplesPerSecond = 5469.0
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0966s; samplesPerSecond = 6624.2
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.1007s; samplesPerSecond = 6355.6
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0970s; samplesPerSecond = 6595.0
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0989s; samplesPerSecond = 6474.5
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0895s; samplesPerSecond = 7151.2
|
||||
MPI Rank 0: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0982s; samplesPerSecond = 6515.7
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.1044s; samplesPerSecond = 6130.0
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0895s; samplesPerSecond = 7150.4
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.1063s; samplesPerSecond = 6020.5
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.1047s; samplesPerSecond = 6114.2
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0983s; samplesPerSecond = 6508.9
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1114s; samplesPerSecond = 5745.5
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0919s; samplesPerSecond = 6961.3
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0783s; samplesPerSecond = 8171.8
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0976s; samplesPerSecond = 6558.1
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6469.9
|
||||
MPI Rank 0: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1018s; samplesPerSecond = 6287.0
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0913s; samplesPerSecond = 7012.0
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.1004s; samplesPerSecond = 6375.1
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.1030s; samplesPerSecond = 6212.9
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1058s; samplesPerSecond = 6051.7
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.1072s; samplesPerSecond = 5972.8
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0980s; samplesPerSecond = 6529.9
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0954s; samplesPerSecond = 6710.5
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0957s; samplesPerSecond = 6684.5
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0673s; samplesPerSecond = 9509.2
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.12839s
|
||||
MPI Rank 0: 08/16/2016 03:20:53: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:53: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -625,45 +625,45 @@ MPI Rank 0: 08/16/2016 03:20:53: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.021385
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.006373
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.017647
|
||||
MPI Rank 0: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalErrorPrediction = 0.61111111 * 2304; time = 0.1981s; samplesPerSecond = 11632.4
|
||||
MPI Rank 0: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalClassificationError = 0.61111111 * 2304; time = 0.1981s; samplesPerSecond = 11632.4
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.009233
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.018749
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.000877
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.016851
|
||||
MPI Rank 0: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.1783s; samplesPerSecond = 14361.8
|
||||
MPI Rank 0: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.1783s; samplesPerSecond = 14361.8
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.004752
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.016704
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00476
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.01635
|
||||
MPI Rank 0: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalErrorPrediction = 0.57617188 * 2560; time = 0.1731s; samplesPerSecond = 14792.0
|
||||
MPI Rank 0: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalClassificationError = 0.57617188 * 2560; time = 0.1731s; samplesPerSecond = 14792.0
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.004797
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.01733
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.007635
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.017449
|
||||
MPI Rank 0: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalErrorPrediction = 0.60664063 * 2560; time = 0.1734s; samplesPerSecond = 14762.2
|
||||
MPI Rank 0: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalClassificationError = 0.60664063 * 2560; time = 0.1734s; samplesPerSecond = 14762.2
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.005846
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.017083
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.004504
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.016877
|
||||
MPI Rank 0: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalErrorPrediction = 0.58945313 * 2560; time = 0.1711s; samplesPerSecond = 14959.0
|
||||
MPI Rank 0: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalClassificationError = 0.58945313 * 2560; time = 0.1711s; samplesPerSecond = 14959.0
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.005357
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.016769
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.007533
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.016959
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalErrorPrediction = 0.56562500 * 2560; time = 0.1716s; samplesPerSecond = 14915.7
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalClassificationError = 0.56562500 * 2560; time = 0.1716s; samplesPerSecond = 14915.7
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00462
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.015866
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00912
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.017219
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalErrorPrediction = 0.59179688 * 2560; time = 0.1732s; samplesPerSecond = 14777.4
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalClassificationError = 0.59179688 * 2560; time = 0.1732s; samplesPerSecond = 14777.4
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.007759
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.016693
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.008586
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.016108
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalErrorPrediction = 0.58632812 * 2560; time = 0.1678s; samplesPerSecond = 15254.6
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalClassificationError = 0.58632812 * 2560; time = 0.1678s; samplesPerSecond = 15254.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.006567
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.006809
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalErrorPrediction = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42703s
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalClassificationError = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42703s
|
||||
MPI Rank 0: 08/16/2016 03:20:55: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -674,13 +674,13 @@ MPI Rank 0: Async gradient aggregation wait time: 0.014735
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.03433
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.004733
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.029133
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalErrorPrediction = 0.56510417 * 9216; time = 0.3403s; samplesPerSecond = 27084.1
|
||||
MPI Rank 0: 08/16/2016 03:20:55: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalClassificationError = 0.56510417 * 9216; time = 0.3403s; samplesPerSecond = 27084.1
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.015933
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.030693
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.015901
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.028981
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalErrorPrediction = 0.56914062 * 10240; time = 0.3115s; samplesPerSecond = 32875.8
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalErrorPrediction = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.677633s
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalClassificationError = 0.56914062 * 10240; time = 0.3115s; samplesPerSecond = 32875.8
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalClassificationError = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.677633s
|
||||
MPI Rank 0: 08/16/2016 03:20:56: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -691,14 +691,14 @@ MPI Rank 0: Async gradient aggregation wait time: 0.004776
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.028351
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.008151
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.028006
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalErrorPrediction = 0.53982205 * 9216; time = 0.3011s; samplesPerSecond = 30607.6
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalClassificationError = 0.53982205 * 9216; time = 0.3011s; samplesPerSecond = 30607.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.014923
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.029642
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.005029
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.028009
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalErrorPrediction = 0.53369141 * 10240; time = 0.2962s; samplesPerSecond = 34566.3
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalClassificationError = 0.53369141 * 10240; time = 0.2962s; samplesPerSecond = 34566.3
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00696
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalErrorPrediction = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=0.616271s
|
||||
MPI Rank 0: 08/16/2016 03:20:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalClassificationError = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=0.616271s
|
||||
MPI Rank 0: 08/16/2016 03:20:56: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816031852.202534\Speech\DNN_ParallelBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:20:56: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -752,7 +752,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -778,7 +778,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -861,7 +861,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -887,7 +887,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -976,7 +976,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1002,7 +1002,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1099,7 +1099,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1128,7 +1128,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1151,7 +1151,7 @@ MPI Rank 1: 08/16/2016 03:20:45: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:20:45: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:45: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:20:45: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:20:45: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1204,39 +1204,39 @@ MPI Rank 1: 08/16/2016 03:20:50: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:50: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1258s; samplesPerSecond = 5085.8
|
||||
MPI Rank 1: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.1021s; samplesPerSecond = 6269.4
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0950s; samplesPerSecond = 6739.7
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1003s; samplesPerSecond = 6378.9
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0998s; samplesPerSecond = 6411.5
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0817s; samplesPerSecond = 7830.8
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.1018s; samplesPerSecond = 6285.0
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.1112s; samplesPerSecond = 5754.3
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.1121s; samplesPerSecond = 5706.8
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.1090s; samplesPerSecond = 5870.2
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0955s; samplesPerSecond = 6702.7
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.1005s; samplesPerSecond = 6370.6
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.1134s; samplesPerSecond = 5644.0
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0977s; samplesPerSecond = 6548.3
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0888s; samplesPerSecond = 7204.6
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.1079s; samplesPerSecond = 5931.9
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0987s; samplesPerSecond = 6483.4
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0913s; samplesPerSecond = 7009.7
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.1129s; samplesPerSecond = 5668.2
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0997s; samplesPerSecond = 6422.2
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0989s; samplesPerSecond = 6472.4
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1053s; samplesPerSecond = 6077.0
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1044s; samplesPerSecond = 6129.0
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0970s; samplesPerSecond = 6597.4
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0917s; samplesPerSecond = 6982.9
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.1018s; samplesPerSecond = 6289.2
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0895s; samplesPerSecond = 7150.7
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0979s; samplesPerSecond = 6540.5
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0984s; samplesPerSecond = 6502.9
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.1055s; samplesPerSecond = 6066.1
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0707s; samplesPerSecond = 9058.1
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0399s; samplesPerSecond = 16044.1
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16512s
|
||||
MPI Rank 1: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1258s; samplesPerSecond = 5085.8
|
||||
MPI Rank 1: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.1021s; samplesPerSecond = 6269.4
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0950s; samplesPerSecond = 6739.7
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1003s; samplesPerSecond = 6378.9
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0998s; samplesPerSecond = 6411.5
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0817s; samplesPerSecond = 7830.8
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.1018s; samplesPerSecond = 6285.0
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.1112s; samplesPerSecond = 5754.3
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.1121s; samplesPerSecond = 5706.8
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.1090s; samplesPerSecond = 5870.2
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0955s; samplesPerSecond = 6702.7
|
||||
MPI Rank 1: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.1005s; samplesPerSecond = 6370.6
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.1134s; samplesPerSecond = 5644.0
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0977s; samplesPerSecond = 6548.3
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0888s; samplesPerSecond = 7204.6
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.1079s; samplesPerSecond = 5931.9
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0987s; samplesPerSecond = 6483.4
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0913s; samplesPerSecond = 7009.7
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.1129s; samplesPerSecond = 5668.2
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0997s; samplesPerSecond = 6422.2
|
||||
MPI Rank 1: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0989s; samplesPerSecond = 6472.4
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1053s; samplesPerSecond = 6077.0
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1044s; samplesPerSecond = 6129.0
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0970s; samplesPerSecond = 6597.4
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0917s; samplesPerSecond = 6982.9
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.1018s; samplesPerSecond = 6289.2
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0895s; samplesPerSecond = 7150.7
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0979s; samplesPerSecond = 6540.5
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0984s; samplesPerSecond = 6502.9
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.1055s; samplesPerSecond = 6066.1
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0707s; samplesPerSecond = 9058.1
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0399s; samplesPerSecond = 16044.1
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16512s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:53: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1245,45 +1245,45 @@ MPI Rank 1: 08/16/2016 03:20:53: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.016814
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004995
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.018553
|
||||
MPI Rank 1: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalErrorPrediction = 0.61111111 * 2304; time = 0.1969s; samplesPerSecond = 11701.9
|
||||
MPI Rank 1: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalClassificationError = 0.61111111 * 2304; time = 0.1969s; samplesPerSecond = 11701.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.006765
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.017839
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.012349
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.016538
|
||||
MPI Rank 1: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.1777s; samplesPerSecond = 14408.1
|
||||
MPI Rank 1: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.1777s; samplesPerSecond = 14408.1
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.0074
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.016417
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.011519
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.016373
|
||||
MPI Rank 1: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalErrorPrediction = 0.57617188 * 2560; time = 0.1734s; samplesPerSecond = 14765.1
|
||||
MPI Rank 1: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalClassificationError = 0.57617188 * 2560; time = 0.1734s; samplesPerSecond = 14765.1
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.008701
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.017605
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003101
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.017339
|
||||
MPI Rank 1: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalErrorPrediction = 0.60664063 * 2560; time = 0.1734s; samplesPerSecond = 14766.9
|
||||
MPI Rank 1: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalClassificationError = 0.60664063 * 2560; time = 0.1734s; samplesPerSecond = 14766.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007955
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.017248
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.006691
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.016928
|
||||
MPI Rank 1: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalErrorPrediction = 0.58945313 * 2560; time = 0.1709s; samplesPerSecond = 14979.9
|
||||
MPI Rank 1: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalClassificationError = 0.58945313 * 2560; time = 0.1709s; samplesPerSecond = 14979.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004503
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.016657
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.011371
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.017292
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalErrorPrediction = 0.56562500 * 2560; time = 0.1717s; samplesPerSecond = 14909.0
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalClassificationError = 0.56562500 * 2560; time = 0.1717s; samplesPerSecond = 14909.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.008425
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.015857
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004495
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.017283
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalErrorPrediction = 0.59179688 * 2560; time = 0.1735s; samplesPerSecond = 14752.4
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalClassificationError = 0.59179688 * 2560; time = 0.1735s; samplesPerSecond = 14752.4
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.002782
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.017143
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.006964
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.016254
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalErrorPrediction = 0.58632812 * 2560; time = 0.1672s; samplesPerSecond = 15314.9
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalClassificationError = 0.58632812 * 2560; time = 0.1672s; samplesPerSecond = 15314.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007214
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.006841
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalErrorPrediction = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42722s
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalClassificationError = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42722s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1293,13 +1293,13 @@ MPI Rank 1: Async gradient aggregation wait time: 0.010824
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.034649
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.018618
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.02933
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalErrorPrediction = 0.56510417 * 9216; time = 0.3374s; samplesPerSecond = 27316.2
|
||||
MPI Rank 1: 08/16/2016 03:20:55: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalClassificationError = 0.56510417 * 9216; time = 0.3374s; samplesPerSecond = 27316.2
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007969
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.030869
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003987
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.031625
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalErrorPrediction = 0.56914062 * 10240; time = 0.3192s; samplesPerSecond = 32083.2
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalErrorPrediction = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.678136s
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalClassificationError = 0.56914062 * 10240; time = 0.3192s; samplesPerSecond = 32083.2
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalClassificationError = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.678136s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1309,14 +1309,14 @@ MPI Rank 1: Async gradient aggregation wait time: 0.006331
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.028676
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007827
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.028017
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalErrorPrediction = 0.53982205 * 9216; time = 0.2977s; samplesPerSecond = 30954.9
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalClassificationError = 0.53982205 * 9216; time = 0.2977s; samplesPerSecond = 30954.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007729
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.029894
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.016801
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.028078
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalErrorPrediction = 0.53369141 * 10240; time = 0.2965s; samplesPerSecond = 34541.1
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalClassificationError = 0.53369141 * 10240; time = 0.2965s; samplesPerSecond = 34541.1
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.006883
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalErrorPrediction = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=0.616484s
|
||||
MPI Rank 1: 08/16/2016 03:20:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalClassificationError = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=0.616484s
|
||||
MPI Rank 1: 08/16/2016 03:20:57: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:20:57: Action "train" complete.
|
||||
|
|
@ -1369,7 +1369,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1395,7 +1395,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1478,7 +1478,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1504,7 +1504,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1593,7 +1593,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1619,7 +1619,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1716,7 +1716,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1745,7 +1745,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1768,7 +1768,7 @@ MPI Rank 2: 08/16/2016 03:20:44: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 03:20:44: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:44: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 03:20:44: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 03:20:44: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1821,39 +1821,39 @@ MPI Rank 2: 08/16/2016 03:20:50: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:50: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1040s; samplesPerSecond = 6155.6
|
||||
MPI Rank 2: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0991s; samplesPerSecond = 6456.2
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0984s; samplesPerSecond = 6500.9
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0972s; samplesPerSecond = 6587.3
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0920s; samplesPerSecond = 6953.0
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0921s; samplesPerSecond = 6946.9
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0988s; samplesPerSecond = 6476.8
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.0920s; samplesPerSecond = 6958.0
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0844s; samplesPerSecond = 7583.0
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0780s; samplesPerSecond = 8202.9
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0968s; samplesPerSecond = 6608.2
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.1046s; samplesPerSecond = 6118.0
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0919s; samplesPerSecond = 6960.6
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0943s; samplesPerSecond = 6788.1
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0877s; samplesPerSecond = 7298.6
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0951s; samplesPerSecond = 6726.9
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0914s; samplesPerSecond = 7002.0
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0951s; samplesPerSecond = 6726.4
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0921s; samplesPerSecond = 6952.3
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0981s; samplesPerSecond = 6525.8
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1111s; samplesPerSecond = 5763.0
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0984s; samplesPerSecond = 6506.5
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0886s; samplesPerSecond = 7225.4
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0954s; samplesPerSecond = 6710.3
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0975s; samplesPerSecond = 6562.5
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.1008s; samplesPerSecond = 6349.5
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0910s; samplesPerSecond = 7033.9
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0879s; samplesPerSecond = 7278.0
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.1004s; samplesPerSecond = 6372.2
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0976s; samplesPerSecond = 6560.3
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0959s; samplesPerSecond = 6677.0
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0958s; samplesPerSecond = 6683.8
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.06077s
|
||||
MPI Rank 2: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1040s; samplesPerSecond = 6155.6
|
||||
MPI Rank 2: 08/16/2016 03:20:50: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0991s; samplesPerSecond = 6456.2
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0984s; samplesPerSecond = 6500.9
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0972s; samplesPerSecond = 6587.3
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0920s; samplesPerSecond = 6953.0
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0921s; samplesPerSecond = 6946.9
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0988s; samplesPerSecond = 6476.8
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.0920s; samplesPerSecond = 6958.0
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0844s; samplesPerSecond = 7583.0
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0780s; samplesPerSecond = 8202.9
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0968s; samplesPerSecond = 6608.2
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.1046s; samplesPerSecond = 6118.0
|
||||
MPI Rank 2: 08/16/2016 03:20:51: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0919s; samplesPerSecond = 6960.6
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0943s; samplesPerSecond = 6788.1
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0877s; samplesPerSecond = 7298.6
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0951s; samplesPerSecond = 6726.9
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0914s; samplesPerSecond = 7002.0
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0951s; samplesPerSecond = 6726.4
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0921s; samplesPerSecond = 6952.3
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0981s; samplesPerSecond = 6525.8
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1111s; samplesPerSecond = 5763.0
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0984s; samplesPerSecond = 6506.5
|
||||
MPI Rank 2: 08/16/2016 03:20:52: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0886s; samplesPerSecond = 7225.4
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0954s; samplesPerSecond = 6710.3
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0975s; samplesPerSecond = 6562.5
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.1008s; samplesPerSecond = 6349.5
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0910s; samplesPerSecond = 7033.9
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0879s; samplesPerSecond = 7278.0
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.1004s; samplesPerSecond = 6372.2
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0976s; samplesPerSecond = 6560.3
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0959s; samplesPerSecond = 6677.0
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0958s; samplesPerSecond = 6683.8
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.06077s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:53: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1862,45 +1862,45 @@ MPI Rank 2: 08/16/2016 03:20:53: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.035327
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.00284
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.018497
|
||||
MPI Rank 2: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalErrorPrediction = 0.61111111 * 2304; time = 0.2022s; samplesPerSecond = 11394.7
|
||||
MPI Rank 2: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.22369214 * 2304; EvalClassificationError = 0.61111111 * 2304; time = 0.2022s; samplesPerSecond = 11394.7
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.01201
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.013216
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.016512
|
||||
MPI Rank 2: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.1741s; samplesPerSecond = 14705.9
|
||||
MPI Rank 2: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.23347641 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.1741s; samplesPerSecond = 14705.9
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.010019
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.016346
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.008387
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.016248
|
||||
MPI Rank 2: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalErrorPrediction = 0.57617188 * 2560; time = 0.1733s; samplesPerSecond = 14771.6
|
||||
MPI Rank 2: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.16589380 * 2560; EvalClassificationError = 0.57617188 * 2560; time = 0.1733s; samplesPerSecond = 14771.6
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.00665
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.017602
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.006178
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.017222
|
||||
MPI Rank 2: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalErrorPrediction = 0.60664063 * 2560; time = 0.1735s; samplesPerSecond = 14758.8
|
||||
MPI Rank 2: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.17067204 * 2560; EvalClassificationError = 0.60664063 * 2560; time = 0.1735s; samplesPerSecond = 14758.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004815
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.016551
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.009325
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.016836
|
||||
MPI Rank 2: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalErrorPrediction = 0.58945313 * 2560; time = 0.1709s; samplesPerSecond = 14982.6
|
||||
MPI Rank 2: 08/16/2016 03:20:54: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.18189249 * 2560; EvalClassificationError = 0.58945313 * 2560; time = 0.1709s; samplesPerSecond = 14982.6
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.005557
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.016515
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.000651
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.017232
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalErrorPrediction = 0.56562500 * 2560; time = 0.1717s; samplesPerSecond = 14910.2
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.08724350 * 2560; EvalClassificationError = 0.56562500 * 2560; time = 0.1717s; samplesPerSecond = 14910.2
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004326
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.016551
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.005588
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.017189
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalErrorPrediction = 0.59179688 * 2560; time = 0.1735s; samplesPerSecond = 14752.5
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.09218108 * 2560; EvalClassificationError = 0.59179688 * 2560; time = 0.1735s; samplesPerSecond = 14752.5
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004491
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.016498
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.003218
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.016326
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalErrorPrediction = 0.58632812 * 2560; time = 0.1679s; samplesPerSecond = 15250.3
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10191157 * 2560; EvalClassificationError = 0.58632812 * 2560; time = 0.1679s; samplesPerSecond = 15250.3
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.006218
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.006778
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalErrorPrediction = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42687s
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15620068 * 20480; EvalClassificationError = 0.58857422 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.42687s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1910,13 +1910,13 @@ MPI Rank 2: Async gradient aggregation wait time: 1e-006
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.016322
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.013477
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.029454
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalErrorPrediction = 0.56510417 * 9216; time = 0.3412s; samplesPerSecond = 27012.3
|
||||
MPI Rank 2: 08/16/2016 03:20:55: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11767394 * 9216; EvalClassificationError = 0.56510417 * 9216; time = 0.3412s; samplesPerSecond = 27012.3
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004206
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.03017
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.009311
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.03402
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalErrorPrediction = 0.56914062 * 10240; time = 0.3164s; samplesPerSecond = 32364.4
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalErrorPrediction = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.677824s
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.08282316 * 10240; EvalClassificationError = 0.56914062 * 10240; time = 0.3164s; samplesPerSecond = 32364.4
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09667113 * 20480; EvalClassificationError = 0.56689453 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.677824s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1926,14 +1926,14 @@ MPI Rank 2: Async gradient aggregation wait time: 0.024966
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.028835
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.002866
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.027712
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalErrorPrediction = 0.53982205 * 9216; time = 0.3015s; samplesPerSecond = 30568.5
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.98353176 * 9216; EvalClassificationError = 0.53982205 * 9216; time = 0.3015s; samplesPerSecond = 30568.5
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.003038
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.029717
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.009962
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.028025
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalErrorPrediction = 0.53369141 * 10240; time = 0.2965s; samplesPerSecond = 34531.1
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96715780 * 10240; EvalClassificationError = 0.53369141 * 10240; time = 0.2965s; samplesPerSecond = 34531.1
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.006611
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalErrorPrediction = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=0.616116s
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.97591825 * 20480; EvalClassificationError = 0.53642578 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=0.616116s
|
||||
MPI Rank 2: 08/16/2016 03:20:56: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:20:56: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -105,7 +105,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -131,7 +131,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -228,7 +228,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -254,7 +254,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -358,7 +358,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -384,7 +384,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -505,7 +505,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -534,7 +534,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -557,7 +557,7 @@ MPI Rank 0: 08/16/2016 09:57:51: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 09:57:51: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:57:51: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 09:57:51: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 09:57:51: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -612,73 +612,73 @@ MPI Rank 0: 08/16/2016 09:57:53: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:57:53: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 09:57:53: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.2812s; samplesPerSecond = 2275.8
|
||||
MPI Rank 0: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.2752s; samplesPerSecond = 2325.6
|
||||
MPI Rank 0: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.2728s; samplesPerSecond = 2346.5
|
||||
MPI Rank 0: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.2894s; samplesPerSecond = 2211.3
|
||||
MPI Rank 0: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.2710s; samplesPerSecond = 2361.9
|
||||
MPI Rank 0: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.2699s; samplesPerSecond = 2371.6
|
||||
MPI Rank 0: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.2742s; samplesPerSecond = 2334.4
|
||||
MPI Rank 0: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.2686s; samplesPerSecond = 2382.7
|
||||
MPI Rank 0: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.2677s; samplesPerSecond = 2390.8
|
||||
MPI Rank 0: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.2803s; samplesPerSecond = 2283.2
|
||||
MPI Rank 0: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.2757s; samplesPerSecond = 2321.7
|
||||
MPI Rank 0: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.2605s; samplesPerSecond = 2456.7
|
||||
MPI Rank 0: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.2691s; samplesPerSecond = 2378.1
|
||||
MPI Rank 0: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.2569s; samplesPerSecond = 2491.1
|
||||
MPI Rank 0: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.2695s; samplesPerSecond = 2375.0
|
||||
MPI Rank 0: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.2668s; samplesPerSecond = 2398.8
|
||||
MPI Rank 0: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.2800s; samplesPerSecond = 2285.7
|
||||
MPI Rank 0: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.2769s; samplesPerSecond = 2311.3
|
||||
MPI Rank 0: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.2753s; samplesPerSecond = 2324.9
|
||||
MPI Rank 0: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.2693s; samplesPerSecond = 2376.7
|
||||
MPI Rank 0: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.2769s; samplesPerSecond = 2311.0
|
||||
MPI Rank 0: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.2761s; samplesPerSecond = 2317.6
|
||||
MPI Rank 0: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2747s; samplesPerSecond = 2329.9
|
||||
MPI Rank 0: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2589s; samplesPerSecond = 2472.2
|
||||
MPI Rank 0: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.2558s; samplesPerSecond = 2501.7
|
||||
MPI Rank 0: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.2781s; samplesPerSecond = 2301.0
|
||||
MPI Rank 0: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.2725s; samplesPerSecond = 2348.2
|
||||
MPI Rank 0: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.2654s; samplesPerSecond = 2411.7
|
||||
MPI Rank 0: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.2706s; samplesPerSecond = 2365.1
|
||||
MPI Rank 0: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.2666s; samplesPerSecond = 2400.3
|
||||
MPI Rank 0: 08/16/2016 09:58:02: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.2704s; samplesPerSecond = 2367.0
|
||||
MPI Rank 0: 08/16/2016 09:58:02: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.2689s; samplesPerSecond = 2379.8
|
||||
MPI Rank 0: 08/16/2016 09:58:02: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=8.69276s
|
||||
MPI Rank 0: 08/16/2016 09:57:53: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.2812s; samplesPerSecond = 2275.8
|
||||
MPI Rank 0: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.2752s; samplesPerSecond = 2325.6
|
||||
MPI Rank 0: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.2728s; samplesPerSecond = 2346.5
|
||||
MPI Rank 0: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.2894s; samplesPerSecond = 2211.3
|
||||
MPI Rank 0: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.2710s; samplesPerSecond = 2361.9
|
||||
MPI Rank 0: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.2699s; samplesPerSecond = 2371.6
|
||||
MPI Rank 0: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.2742s; samplesPerSecond = 2334.4
|
||||
MPI Rank 0: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.2686s; samplesPerSecond = 2382.7
|
||||
MPI Rank 0: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.2677s; samplesPerSecond = 2390.8
|
||||
MPI Rank 0: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.2803s; samplesPerSecond = 2283.2
|
||||
MPI Rank 0: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.2757s; samplesPerSecond = 2321.7
|
||||
MPI Rank 0: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.2605s; samplesPerSecond = 2456.7
|
||||
MPI Rank 0: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.2691s; samplesPerSecond = 2378.1
|
||||
MPI Rank 0: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.2569s; samplesPerSecond = 2491.1
|
||||
MPI Rank 0: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.2695s; samplesPerSecond = 2375.0
|
||||
MPI Rank 0: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.2668s; samplesPerSecond = 2398.8
|
||||
MPI Rank 0: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.2800s; samplesPerSecond = 2285.7
|
||||
MPI Rank 0: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.2769s; samplesPerSecond = 2311.3
|
||||
MPI Rank 0: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.2753s; samplesPerSecond = 2324.9
|
||||
MPI Rank 0: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.2693s; samplesPerSecond = 2376.7
|
||||
MPI Rank 0: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.2769s; samplesPerSecond = 2311.0
|
||||
MPI Rank 0: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.2761s; samplesPerSecond = 2317.6
|
||||
MPI Rank 0: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2747s; samplesPerSecond = 2329.9
|
||||
MPI Rank 0: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2589s; samplesPerSecond = 2472.2
|
||||
MPI Rank 0: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.2558s; samplesPerSecond = 2501.7
|
||||
MPI Rank 0: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.2781s; samplesPerSecond = 2301.0
|
||||
MPI Rank 0: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.2725s; samplesPerSecond = 2348.2
|
||||
MPI Rank 0: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.2654s; samplesPerSecond = 2411.7
|
||||
MPI Rank 0: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.2706s; samplesPerSecond = 2365.1
|
||||
MPI Rank 0: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.2666s; samplesPerSecond = 2400.3
|
||||
MPI Rank 0: 08/16/2016 09:58:02: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.2704s; samplesPerSecond = 2367.0
|
||||
MPI Rank 0: 08/16/2016 09:58:02: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.2689s; samplesPerSecond = 2379.8
|
||||
MPI Rank 0: 08/16/2016 09:58:02: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=8.69276s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 10.0 ms
|
||||
MPI Rank 0: 08/16/2016 09:58:31: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.23112813 * 83050; perplexity = 9.31036343; EvalErrorPrediction = 0.61196869 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:58:31: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.23112813 * 83050; EvalErrorPrediction = 0.61196869 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:58:31: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.23112813 * 83050; perplexity = 9.31036343; EvalClassificationError = 0.61196869 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:58:31: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.23112813 * 83050; EvalClassificationError = 0.61196869 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:58:31: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelCrossValidation@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:58:31: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:58:31: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 09:58:32: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.12679700 * 2560; EvalErrorPrediction = 0.56601563 * 2560; time = 0.8259s; samplesPerSecond = 3099.5
|
||||
MPI Rank 0: 08/16/2016 09:58:32: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.04568504 * 2560; EvalErrorPrediction = 0.55429688 * 2560; time = 0.8085s; samplesPerSecond = 3166.4
|
||||
MPI Rank 0: 08/16/2016 09:58:33: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.02935394 * 2560; EvalErrorPrediction = 0.54570312 * 2560; time = 0.7974s; samplesPerSecond = 3210.4
|
||||
MPI Rank 0: 08/16/2016 09:58:34: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.01533190 * 2560; EvalErrorPrediction = 0.55898437 * 2560; time = 0.7963s; samplesPerSecond = 3214.7
|
||||
MPI Rank 0: 08/16/2016 09:58:35: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.00654444 * 2560; EvalErrorPrediction = 0.54296875 * 2560; time = 0.7881s; samplesPerSecond = 3248.3
|
||||
MPI Rank 0: 08/16/2016 09:58:36: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.94945520 * 2560; EvalErrorPrediction = 0.53359375 * 2560; time = 0.7922s; samplesPerSecond = 3231.5
|
||||
MPI Rank 0: 08/16/2016 09:58:36: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.94218281 * 2560; EvalErrorPrediction = 0.52421875 * 2560; time = 0.7953s; samplesPerSecond = 3219.0
|
||||
MPI Rank 0: 08/16/2016 09:58:37: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.92847361 * 2560; EvalErrorPrediction = 0.52929688 * 2560; time = 0.7827s; samplesPerSecond = 3270.9
|
||||
MPI Rank 0: 08/16/2016 09:58:37: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.00547799 * 20480; EvalErrorPrediction = 0.54438477 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=6.39217s
|
||||
MPI Rank 0: 08/16/2016 09:58:32: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.12679700 * 2560; EvalClassificationError = 0.56601563 * 2560; time = 0.8259s; samplesPerSecond = 3099.5
|
||||
MPI Rank 0: 08/16/2016 09:58:32: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.04568504 * 2560; EvalClassificationError = 0.55429688 * 2560; time = 0.8085s; samplesPerSecond = 3166.4
|
||||
MPI Rank 0: 08/16/2016 09:58:33: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.02935394 * 2560; EvalClassificationError = 0.54570312 * 2560; time = 0.7974s; samplesPerSecond = 3210.4
|
||||
MPI Rank 0: 08/16/2016 09:58:34: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.01533190 * 2560; EvalClassificationError = 0.55898437 * 2560; time = 0.7963s; samplesPerSecond = 3214.7
|
||||
MPI Rank 0: 08/16/2016 09:58:35: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.00654444 * 2560; EvalClassificationError = 0.54296875 * 2560; time = 0.7881s; samplesPerSecond = 3248.3
|
||||
MPI Rank 0: 08/16/2016 09:58:36: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.94945520 * 2560; EvalClassificationError = 0.53359375 * 2560; time = 0.7922s; samplesPerSecond = 3231.5
|
||||
MPI Rank 0: 08/16/2016 09:58:36: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.94218281 * 2560; EvalClassificationError = 0.52421875 * 2560; time = 0.7953s; samplesPerSecond = 3219.0
|
||||
MPI Rank 0: 08/16/2016 09:58:37: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.92847361 * 2560; EvalClassificationError = 0.52929688 * 2560; time = 0.7827s; samplesPerSecond = 3270.9
|
||||
MPI Rank 0: 08/16/2016 09:58:37: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.00547799 * 20480; EvalClassificationError = 0.54438477 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=6.39217s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: 08/16/2016 09:59:05: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.91178842 * 83050; perplexity = 6.76517694; EvalErrorPrediction = 0.52972908 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:05: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.91178842 * 83050; EvalErrorPrediction = 0.52972908 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:05: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.91178842 * 83050; perplexity = 6.76517694; EvalClassificationError = 0.52972908 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:05: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.91178842 * 83050; EvalClassificationError = 0.52972908 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:05: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelCrossValidation@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:05: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:05: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.93108721 * 10240; EvalErrorPrediction = 0.52958984 * 10240; time = 2.8844s; samplesPerSecond = 3550.1
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95098710 * 10240; EvalErrorPrediction = 0.54755859 * 10240; time = 2.8076s; samplesPerSecond = 3647.2
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.94103716 * 20480; EvalErrorPrediction = 0.53857422 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=5.70009s
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.93108721 * 10240; EvalClassificationError = 0.52958984 * 10240; time = 2.8844s; samplesPerSecond = 3550.1
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95098710 * 10240; EvalClassificationError = 0.54755859 * 10240; time = 2.8076s; samplesPerSecond = 3647.2
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.94103716 * 20480; EvalClassificationError = 0.53857422 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=5.70009s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: 08/16/2016 09:59:37: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.88963921 * 83050; perplexity = 6.61698094; EvalErrorPrediction = 0.52007225 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:37: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.88963921 * 83050; EvalErrorPrediction = 0.52007225 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:37: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.88963921 * 83050; perplexity = 6.61698094; EvalClassificationError = 0.52007225 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:37: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.88963921 * 83050; EvalClassificationError = 0.52007225 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:37: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelCrossValidation@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 09:59:37: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -731,7 +731,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -757,7 +757,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -854,7 +854,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -880,7 +880,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -984,7 +984,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1010,7 +1010,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1131,7 +1131,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1160,7 +1160,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1183,7 +1183,7 @@ MPI Rank 1: 08/16/2016 09:57:51: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 09:57:51: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:57:51: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 09:57:51: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 09:57:51: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1238,70 +1238,70 @@ MPI Rank 1: 08/16/2016 09:57:53: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:57:53: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 09:57:53: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.2811s; samplesPerSecond = 2276.9
|
||||
MPI Rank 1: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.2753s; samplesPerSecond = 2325.2
|
||||
MPI Rank 1: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.2721s; samplesPerSecond = 2352.1
|
||||
MPI Rank 1: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.2899s; samplesPerSecond = 2207.7
|
||||
MPI Rank 1: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.2708s; samplesPerSecond = 2363.1
|
||||
MPI Rank 1: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.2697s; samplesPerSecond = 2372.6
|
||||
MPI Rank 1: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.2740s; samplesPerSecond = 2335.9
|
||||
MPI Rank 1: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.2688s; samplesPerSecond = 2381.3
|
||||
MPI Rank 1: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.2676s; samplesPerSecond = 2391.6
|
||||
MPI Rank 1: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.2802s; samplesPerSecond = 2284.2
|
||||
MPI Rank 1: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.2758s; samplesPerSecond = 2320.6
|
||||
MPI Rank 1: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.2606s; samplesPerSecond = 2456.1
|
||||
MPI Rank 1: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.2689s; samplesPerSecond = 2380.5
|
||||
MPI Rank 1: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.2573s; samplesPerSecond = 2487.8
|
||||
MPI Rank 1: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.2693s; samplesPerSecond = 2376.7
|
||||
MPI Rank 1: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.2667s; samplesPerSecond = 2399.3
|
||||
MPI Rank 1: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.2795s; samplesPerSecond = 2289.8
|
||||
MPI Rank 1: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.2774s; samplesPerSecond = 2307.5
|
||||
MPI Rank 1: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.2751s; samplesPerSecond = 2326.0
|
||||
MPI Rank 1: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.2692s; samplesPerSecond = 2377.7
|
||||
MPI Rank 1: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.2771s; samplesPerSecond = 2309.9
|
||||
MPI Rank 1: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.2761s; samplesPerSecond = 2318.0
|
||||
MPI Rank 1: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2745s; samplesPerSecond = 2331.9
|
||||
MPI Rank 1: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2589s; samplesPerSecond = 2471.6
|
||||
MPI Rank 1: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.2558s; samplesPerSecond = 2502.1
|
||||
MPI Rank 1: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.2782s; samplesPerSecond = 2300.2
|
||||
MPI Rank 1: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.2723s; samplesPerSecond = 2350.5
|
||||
MPI Rank 1: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.2652s; samplesPerSecond = 2413.6
|
||||
MPI Rank 1: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.2708s; samplesPerSecond = 2363.4
|
||||
MPI Rank 1: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.2668s; samplesPerSecond = 2398.6
|
||||
MPI Rank 1: 08/16/2016 09:58:02: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.2702s; samplesPerSecond = 2368.3
|
||||
MPI Rank 1: 08/16/2016 09:58:02: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.2689s; samplesPerSecond = 2380.4
|
||||
MPI Rank 1: 08/16/2016 09:58:02: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=8.69267s
|
||||
MPI Rank 1: 08/16/2016 09:57:53: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.2811s; samplesPerSecond = 2276.9
|
||||
MPI Rank 1: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.2753s; samplesPerSecond = 2325.2
|
||||
MPI Rank 1: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.2721s; samplesPerSecond = 2352.1
|
||||
MPI Rank 1: 08/16/2016 09:57:54: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.2899s; samplesPerSecond = 2207.7
|
||||
MPI Rank 1: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.2708s; samplesPerSecond = 2363.1
|
||||
MPI Rank 1: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.2697s; samplesPerSecond = 2372.6
|
||||
MPI Rank 1: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.2740s; samplesPerSecond = 2335.9
|
||||
MPI Rank 1: 08/16/2016 09:57:55: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.2688s; samplesPerSecond = 2381.3
|
||||
MPI Rank 1: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.2676s; samplesPerSecond = 2391.6
|
||||
MPI Rank 1: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.2802s; samplesPerSecond = 2284.2
|
||||
MPI Rank 1: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.2758s; samplesPerSecond = 2320.6
|
||||
MPI Rank 1: 08/16/2016 09:57:56: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.2606s; samplesPerSecond = 2456.1
|
||||
MPI Rank 1: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.2689s; samplesPerSecond = 2380.5
|
||||
MPI Rank 1: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.2573s; samplesPerSecond = 2487.8
|
||||
MPI Rank 1: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.2693s; samplesPerSecond = 2376.7
|
||||
MPI Rank 1: 08/16/2016 09:57:57: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.2667s; samplesPerSecond = 2399.3
|
||||
MPI Rank 1: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.2795s; samplesPerSecond = 2289.8
|
||||
MPI Rank 1: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.2774s; samplesPerSecond = 2307.5
|
||||
MPI Rank 1: 08/16/2016 09:57:58: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.2751s; samplesPerSecond = 2326.0
|
||||
MPI Rank 1: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.2692s; samplesPerSecond = 2377.7
|
||||
MPI Rank 1: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.2771s; samplesPerSecond = 2309.9
|
||||
MPI Rank 1: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.2761s; samplesPerSecond = 2318.0
|
||||
MPI Rank 1: 08/16/2016 09:57:59: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2745s; samplesPerSecond = 2331.9
|
||||
MPI Rank 1: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2589s; samplesPerSecond = 2471.6
|
||||
MPI Rank 1: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.2558s; samplesPerSecond = 2502.1
|
||||
MPI Rank 1: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.2782s; samplesPerSecond = 2300.2
|
||||
MPI Rank 1: 08/16/2016 09:58:00: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.2723s; samplesPerSecond = 2350.5
|
||||
MPI Rank 1: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.2652s; samplesPerSecond = 2413.6
|
||||
MPI Rank 1: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.2708s; samplesPerSecond = 2363.4
|
||||
MPI Rank 1: 08/16/2016 09:58:01: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.2668s; samplesPerSecond = 2398.6
|
||||
MPI Rank 1: 08/16/2016 09:58:02: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.2702s; samplesPerSecond = 2368.3
|
||||
MPI Rank 1: 08/16/2016 09:58:02: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.2689s; samplesPerSecond = 2380.4
|
||||
MPI Rank 1: 08/16/2016 09:58:02: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=8.69267s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 09:58:31: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.23112813 * 83050; perplexity = 9.31036343; EvalErrorPrediction = 0.61196869 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:58:31: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.23112813 * 83050; EvalErrorPrediction = 0.61196869 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:58:31: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.23112813 * 83050; perplexity = 9.31036343; EvalClassificationError = 0.61196869 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:58:31: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.23112813 * 83050; EvalClassificationError = 0.61196869 * 83050
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:31: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:31: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 09:58:32: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.12679700 * 2560; EvalErrorPrediction = 0.56601563 * 2560; time = 0.8264s; samplesPerSecond = 3097.6
|
||||
MPI Rank 1: 08/16/2016 09:58:32: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.04568504 * 2560; EvalErrorPrediction = 0.55429688 * 2560; time = 0.8080s; samplesPerSecond = 3168.2
|
||||
MPI Rank 1: 08/16/2016 09:58:33: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.02935394 * 2560; EvalErrorPrediction = 0.54570312 * 2560; time = 0.7977s; samplesPerSecond = 3209.3
|
||||
MPI Rank 1: 08/16/2016 09:58:34: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.01533190 * 2560; EvalErrorPrediction = 0.55898437 * 2560; time = 0.7962s; samplesPerSecond = 3215.1
|
||||
MPI Rank 1: 08/16/2016 09:58:35: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.00654444 * 2560; EvalErrorPrediction = 0.54296875 * 2560; time = 0.7883s; samplesPerSecond = 3247.5
|
||||
MPI Rank 1: 08/16/2016 09:58:36: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.94945520 * 2560; EvalErrorPrediction = 0.53359375 * 2560; time = 0.7919s; samplesPerSecond = 3232.6
|
||||
MPI Rank 1: 08/16/2016 09:58:36: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.94218281 * 2560; EvalErrorPrediction = 0.52421875 * 2560; time = 0.7953s; samplesPerSecond = 3219.0
|
||||
MPI Rank 1: 08/16/2016 09:58:37: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.92847361 * 2560; EvalErrorPrediction = 0.52929688 * 2560; time = 0.7827s; samplesPerSecond = 3270.8
|
||||
MPI Rank 1: 08/16/2016 09:58:37: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.00547799 * 20480; EvalErrorPrediction = 0.54438477 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=6.3921s
|
||||
MPI Rank 1: 08/16/2016 09:58:32: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.12679700 * 2560; EvalClassificationError = 0.56601563 * 2560; time = 0.8264s; samplesPerSecond = 3097.6
|
||||
MPI Rank 1: 08/16/2016 09:58:32: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.04568504 * 2560; EvalClassificationError = 0.55429688 * 2560; time = 0.8080s; samplesPerSecond = 3168.2
|
||||
MPI Rank 1: 08/16/2016 09:58:33: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.02935394 * 2560; EvalClassificationError = 0.54570312 * 2560; time = 0.7977s; samplesPerSecond = 3209.3
|
||||
MPI Rank 1: 08/16/2016 09:58:34: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.01533190 * 2560; EvalClassificationError = 0.55898437 * 2560; time = 0.7962s; samplesPerSecond = 3215.1
|
||||
MPI Rank 1: 08/16/2016 09:58:35: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.00654444 * 2560; EvalClassificationError = 0.54296875 * 2560; time = 0.7883s; samplesPerSecond = 3247.5
|
||||
MPI Rank 1: 08/16/2016 09:58:36: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.94945520 * 2560; EvalClassificationError = 0.53359375 * 2560; time = 0.7919s; samplesPerSecond = 3232.6
|
||||
MPI Rank 1: 08/16/2016 09:58:36: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.94218281 * 2560; EvalClassificationError = 0.52421875 * 2560; time = 0.7953s; samplesPerSecond = 3219.0
|
||||
MPI Rank 1: 08/16/2016 09:58:37: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.92847361 * 2560; EvalClassificationError = 0.52929688 * 2560; time = 0.7827s; samplesPerSecond = 3270.8
|
||||
MPI Rank 1: 08/16/2016 09:58:37: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.00547799 * 20480; EvalClassificationError = 0.54438477 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=6.3921s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 09:59:05: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.91178842 * 83050; perplexity = 6.76517694; EvalErrorPrediction = 0.52972908 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:05: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.91178842 * 83050; EvalErrorPrediction = 0.52972908 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:05: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.91178842 * 83050; perplexity = 6.76517694; EvalClassificationError = 0.52972908 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:05: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.91178842 * 83050; EvalClassificationError = 0.52972908 * 83050
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:05: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:05: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.93108721 * 10240; EvalErrorPrediction = 0.52958984 * 10240; time = 2.8859s; samplesPerSecond = 3548.3
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95098710 * 10240; EvalErrorPrediction = 0.54755859 * 10240; time = 2.8075s; samplesPerSecond = 3647.4
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.94103716 * 20480; EvalErrorPrediction = 0.53857422 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=5.7s
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.93108721 * 10240; EvalClassificationError = 0.52958984 * 10240; time = 2.8859s; samplesPerSecond = 3548.3
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95098710 * 10240; EvalClassificationError = 0.54755859 * 10240; time = 2.8075s; samplesPerSecond = 3647.4
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.94103716 * 20480; EvalClassificationError = 0.53857422 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=5.7s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 09:59:37: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.88963921 * 83050; perplexity = 6.61698094; EvalErrorPrediction = 0.52007225 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:37: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.88963921 * 83050; EvalErrorPrediction = 0.52007225 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:37: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.88963921 * 83050; perplexity = 6.61698094; EvalClassificationError = 0.52007225 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:37: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.88963921 * 83050; EvalClassificationError = 0.52007225 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:37: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:37: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -105,7 +105,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -131,7 +131,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -228,7 +228,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -254,7 +254,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -358,7 +358,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -384,7 +384,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -506,7 +506,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -535,7 +535,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -558,7 +558,7 @@ MPI Rank 0: 08/16/2016 09:59:41: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 09:59:41: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:41: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 09:59:41: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 09:59:41: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -613,73 +613,73 @@ MPI Rank 0: 08/16/2016 09:59:43: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:43: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 09:59:43: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1080s; samplesPerSecond = 5925.3
|
||||
MPI Rank 0: 08/16/2016 09:59:43: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.1000s; samplesPerSecond = 6397.8
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.1013s; samplesPerSecond = 6319.2
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0994s; samplesPerSecond = 6437.1
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0999s; samplesPerSecond = 6404.9
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.1008s; samplesPerSecond = 6348.5
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.1001s; samplesPerSecond = 6390.9
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.1022s; samplesPerSecond = 6262.2
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.1031s; samplesPerSecond = 6210.3
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0992s; samplesPerSecond = 6454.5
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0998s; samplesPerSecond = 6412.1
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.1006s; samplesPerSecond = 6361.3
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0995s; samplesPerSecond = 6430.6
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.1003s; samplesPerSecond = 6379.9
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0999s; samplesPerSecond = 6404.0
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0992s; samplesPerSecond = 6454.3
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.1004s; samplesPerSecond = 6375.9
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0994s; samplesPerSecond = 6435.8
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.1004s; samplesPerSecond = 6376.7
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0998s; samplesPerSecond = 6413.0
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1008s; samplesPerSecond = 6351.2
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0979s; samplesPerSecond = 6538.0
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0998s; samplesPerSecond = 6415.3
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1005s; samplesPerSecond = 6368.9
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0999s; samplesPerSecond = 6405.0
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0996s; samplesPerSecond = 6427.6
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0997s; samplesPerSecond = 6417.7
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.1002s; samplesPerSecond = 6387.7
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.1008s; samplesPerSecond = 6351.5
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.1003s; samplesPerSecond = 6382.4
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.1001s; samplesPerSecond = 6396.3
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0991s; samplesPerSecond = 6456.1
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.22028s
|
||||
MPI Rank 0: 08/16/2016 09:59:43: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1080s; samplesPerSecond = 5925.3
|
||||
MPI Rank 0: 08/16/2016 09:59:43: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.1000s; samplesPerSecond = 6397.8
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.1013s; samplesPerSecond = 6319.2
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0994s; samplesPerSecond = 6437.1
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0999s; samplesPerSecond = 6404.9
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.1008s; samplesPerSecond = 6348.5
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.1001s; samplesPerSecond = 6390.9
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.1022s; samplesPerSecond = 6262.2
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.1031s; samplesPerSecond = 6210.3
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0992s; samplesPerSecond = 6454.5
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0998s; samplesPerSecond = 6412.1
|
||||
MPI Rank 0: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.1006s; samplesPerSecond = 6361.3
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0995s; samplesPerSecond = 6430.6
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.1003s; samplesPerSecond = 6379.9
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0999s; samplesPerSecond = 6404.0
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0992s; samplesPerSecond = 6454.3
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.1004s; samplesPerSecond = 6375.9
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0994s; samplesPerSecond = 6435.8
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.1004s; samplesPerSecond = 6376.7
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0998s; samplesPerSecond = 6413.0
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1008s; samplesPerSecond = 6351.2
|
||||
MPI Rank 0: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0979s; samplesPerSecond = 6538.0
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0998s; samplesPerSecond = 6415.3
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1005s; samplesPerSecond = 6368.9
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0999s; samplesPerSecond = 6405.0
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0996s; samplesPerSecond = 6427.6
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0997s; samplesPerSecond = 6417.7
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.1002s; samplesPerSecond = 6387.7
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.1008s; samplesPerSecond = 6351.5
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.1003s; samplesPerSecond = 6382.4
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.1001s; samplesPerSecond = 6396.3
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0991s; samplesPerSecond = 6456.1
|
||||
MPI Rank 0: 08/16/2016 09:59:46: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.22028s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 10.0 ms
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.24821047 * 83050; perplexity = 9.47077247; EvalErrorPrediction = 0.61623119 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.24821047 * 83050; EvalErrorPrediction = 0.61623119 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.24821047 * 83050; perplexity = 9.47077247; EvalClassificationError = 0.61623119 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.24821047 * 83050; EvalClassificationError = 0.61623119 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:49: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelCrossValidation@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.13894071 * 2560; EvalErrorPrediction = 0.56992188 * 2560; time = 0.1288s; samplesPerSecond = 19871.6
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.06106261 * 2560; EvalErrorPrediction = 0.55664062 * 2560; time = 0.1249s; samplesPerSecond = 20497.2
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.04459475 * 2560; EvalErrorPrediction = 0.55039063 * 2560; time = 0.1248s; samplesPerSecond = 20516.8
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.03347291 * 2560; EvalErrorPrediction = 0.55742187 * 2560; time = 0.1270s; samplesPerSecond = 20161.6
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02079287 * 2560; EvalErrorPrediction = 0.54414063 * 2560; time = 0.1266s; samplesPerSecond = 20219.3
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96950012 * 2560; EvalErrorPrediction = 0.53085938 * 2560; time = 0.1269s; samplesPerSecond = 20173.8
|
||||
MPI Rank 0: 08/16/2016 09:59:50: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95934862 * 2560; EvalErrorPrediction = 0.52812500 * 2560; time = 0.1253s; samplesPerSecond = 20432.8
|
||||
MPI Rank 0: 08/16/2016 09:59:50: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.94070839 * 2560; EvalErrorPrediction = 0.53125000 * 2560; time = 0.1236s; samplesPerSecond = 20704.9
|
||||
MPI Rank 0: 08/16/2016 09:59:50: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02105262 * 20480; EvalErrorPrediction = 0.54609375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.01496s
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.13894071 * 2560; EvalClassificationError = 0.56992188 * 2560; time = 0.1288s; samplesPerSecond = 19871.6
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.06106261 * 2560; EvalClassificationError = 0.55664062 * 2560; time = 0.1249s; samplesPerSecond = 20497.2
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.04459475 * 2560; EvalClassificationError = 0.55039063 * 2560; time = 0.1248s; samplesPerSecond = 20516.8
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.03347291 * 2560; EvalClassificationError = 0.55742187 * 2560; time = 0.1270s; samplesPerSecond = 20161.6
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02079287 * 2560; EvalClassificationError = 0.54414063 * 2560; time = 0.1266s; samplesPerSecond = 20219.3
|
||||
MPI Rank 0: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96950012 * 2560; EvalClassificationError = 0.53085938 * 2560; time = 0.1269s; samplesPerSecond = 20173.8
|
||||
MPI Rank 0: 08/16/2016 09:59:50: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95934862 * 2560; EvalClassificationError = 0.52812500 * 2560; time = 0.1253s; samplesPerSecond = 20432.8
|
||||
MPI Rank 0: 08/16/2016 09:59:50: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.94070839 * 2560; EvalClassificationError = 0.53125000 * 2560; time = 0.1236s; samplesPerSecond = 20704.9
|
||||
MPI Rank 0: 08/16/2016 09:59:50: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02105262 * 20480; EvalClassificationError = 0.54609375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.01496s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.92733488 * 83050; perplexity = 6.87117331; EvalErrorPrediction = 0.53122216 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.92733488 * 83050; EvalErrorPrediction = 0.53122216 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.92733488 * 83050; perplexity = 6.87117331; EvalClassificationError = 0.53122216 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.92733488 * 83050; EvalClassificationError = 0.53122216 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:51: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelCrossValidation@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94336419 * 10240; EvalErrorPrediction = 0.53056641 * 10240; time = 0.2783s; samplesPerSecond = 36788.6
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96525554 * 10240; EvalErrorPrediction = 0.54873047 * 10240; time = 0.2723s; samplesPerSecond = 37610.7
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95430987 * 20480; EvalErrorPrediction = 0.53964844 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.560004s
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94336419 * 10240; EvalClassificationError = 0.53056641 * 10240; time = 0.2783s; samplesPerSecond = 36788.6
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96525554 * 10240; EvalClassificationError = 0.54873047 * 10240; time = 0.2723s; samplesPerSecond = 37610.7
|
||||
MPI Rank 0: 08/16/2016 09:59:51: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95430987 * 20480; EvalClassificationError = 0.53964844 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.560004s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: 08/16/2016 09:59:52: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.90639119 * 83050; perplexity = 6.72876207; EvalErrorPrediction = 0.52304636 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:52: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.90639119 * 83050; EvalErrorPrediction = 0.52304636 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:52: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.90639119 * 83050; perplexity = 6.72876207; EvalClassificationError = 0.52304636 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:52: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.90639119 * 83050; EvalClassificationError = 0.52304636 * 83050
|
||||
MPI Rank 0: 08/16/2016 09:59:52: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelCrossValidation@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 09:59:52: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -732,7 +732,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -758,7 +758,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -855,7 +855,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -881,7 +881,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -985,7 +985,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1011,7 +1011,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1133,7 +1133,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1162,7 +1162,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1185,7 +1185,7 @@ MPI Rank 1: 08/16/2016 09:59:41: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 09:59:41: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:41: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 09:59:41: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 09:59:41: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1240,70 +1240,70 @@ MPI Rank 1: 08/16/2016 09:59:43: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:43: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 09:59:43: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1079s; samplesPerSecond = 5931.0
|
||||
MPI Rank 1: 08/16/2016 09:59:43: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.1000s; samplesPerSecond = 6401.4
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.1012s; samplesPerSecond = 6325.5
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0994s; samplesPerSecond = 6435.8
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0999s; samplesPerSecond = 6408.6
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.1008s; samplesPerSecond = 6351.6
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.1001s; samplesPerSecond = 6394.2
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.1022s; samplesPerSecond = 6262.5
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.1030s; samplesPerSecond = 6214.2
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0991s; samplesPerSecond = 6457.4
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0998s; samplesPerSecond = 6415.0
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.1005s; samplesPerSecond = 6365.1
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0994s; samplesPerSecond = 6435.5
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.1003s; samplesPerSecond = 6382.7
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0999s; samplesPerSecond = 6407.5
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0991s; samplesPerSecond = 6457.9
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.1003s; samplesPerSecond = 6380.4
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0994s; samplesPerSecond = 6437.5
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.1003s; samplesPerSecond = 6383.1
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0998s; samplesPerSecond = 6411.1
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1007s; samplesPerSecond = 6354.6
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0978s; samplesPerSecond = 6543.0
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0997s; samplesPerSecond = 6416.4
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1004s; samplesPerSecond = 6374.1
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0999s; samplesPerSecond = 6409.4
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0995s; samplesPerSecond = 6429.1
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0997s; samplesPerSecond = 6421.6
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.1001s; samplesPerSecond = 6391.1
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.1007s; samplesPerSecond = 6355.2
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.1002s; samplesPerSecond = 6386.8
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.1000s; samplesPerSecond = 6398.9
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0991s; samplesPerSecond = 6459.2
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.22015s
|
||||
MPI Rank 1: 08/16/2016 09:59:43: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1079s; samplesPerSecond = 5931.0
|
||||
MPI Rank 1: 08/16/2016 09:59:43: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.1000s; samplesPerSecond = 6401.4
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.1012s; samplesPerSecond = 6325.5
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0994s; samplesPerSecond = 6435.8
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0999s; samplesPerSecond = 6408.6
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.1008s; samplesPerSecond = 6351.6
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.1001s; samplesPerSecond = 6394.2
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.1022s; samplesPerSecond = 6262.5
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.1030s; samplesPerSecond = 6214.2
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0991s; samplesPerSecond = 6457.4
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0998s; samplesPerSecond = 6415.0
|
||||
MPI Rank 1: 08/16/2016 09:59:44: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.1005s; samplesPerSecond = 6365.1
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0994s; samplesPerSecond = 6435.5
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.1003s; samplesPerSecond = 6382.7
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0999s; samplesPerSecond = 6407.5
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0991s; samplesPerSecond = 6457.9
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.1003s; samplesPerSecond = 6380.4
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0994s; samplesPerSecond = 6437.5
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.1003s; samplesPerSecond = 6383.1
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0998s; samplesPerSecond = 6411.1
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1007s; samplesPerSecond = 6354.6
|
||||
MPI Rank 1: 08/16/2016 09:59:45: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0978s; samplesPerSecond = 6543.0
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0997s; samplesPerSecond = 6416.4
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1004s; samplesPerSecond = 6374.1
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0999s; samplesPerSecond = 6409.4
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0995s; samplesPerSecond = 6429.1
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0997s; samplesPerSecond = 6421.6
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.1001s; samplesPerSecond = 6391.1
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.1007s; samplesPerSecond = 6355.2
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.1002s; samplesPerSecond = 6386.8
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.1000s; samplesPerSecond = 6398.9
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0991s; samplesPerSecond = 6459.2
|
||||
MPI Rank 1: 08/16/2016 09:59:46: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.22015s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.24821047 * 83050; perplexity = 9.47077247; EvalErrorPrediction = 0.61623119 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.24821047 * 83050; EvalErrorPrediction = 0.61623119 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.24821047 * 83050; perplexity = 9.47077247; EvalClassificationError = 0.61623119 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.24821047 * 83050; EvalClassificationError = 0.61623119 * 83050
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.13894071 * 2560; EvalErrorPrediction = 0.56992188 * 2560; time = 0.1286s; samplesPerSecond = 19904.8
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.06106261 * 2560; EvalErrorPrediction = 0.55664062 * 2560; time = 0.1249s; samplesPerSecond = 20504.1
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.04459475 * 2560; EvalErrorPrediction = 0.55039063 * 2560; time = 0.1247s; samplesPerSecond = 20526.1
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.03347291 * 2560; EvalErrorPrediction = 0.55742187 * 2560; time = 0.1269s; samplesPerSecond = 20170.8
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02079287 * 2560; EvalErrorPrediction = 0.54414063 * 2560; time = 0.1265s; samplesPerSecond = 20242.0
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96950012 * 2560; EvalErrorPrediction = 0.53085938 * 2560; time = 0.1270s; samplesPerSecond = 20163.5
|
||||
MPI Rank 1: 08/16/2016 09:59:50: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95934862 * 2560; EvalErrorPrediction = 0.52812500 * 2560; time = 0.1253s; samplesPerSecond = 20436.0
|
||||
MPI Rank 1: 08/16/2016 09:59:50: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.94070839 * 2560; EvalErrorPrediction = 0.53125000 * 2560; time = 0.1236s; samplesPerSecond = 20712.5
|
||||
MPI Rank 1: 08/16/2016 09:59:50: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02105262 * 20480; EvalErrorPrediction = 0.54609375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.01478s
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.13894071 * 2560; EvalClassificationError = 0.56992188 * 2560; time = 0.1286s; samplesPerSecond = 19904.8
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.06106261 * 2560; EvalClassificationError = 0.55664062 * 2560; time = 0.1249s; samplesPerSecond = 20504.1
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.04459475 * 2560; EvalClassificationError = 0.55039063 * 2560; time = 0.1247s; samplesPerSecond = 20526.1
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.03347291 * 2560; EvalClassificationError = 0.55742187 * 2560; time = 0.1269s; samplesPerSecond = 20170.8
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02079287 * 2560; EvalClassificationError = 0.54414063 * 2560; time = 0.1265s; samplesPerSecond = 20242.0
|
||||
MPI Rank 1: 08/16/2016 09:59:49: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96950012 * 2560; EvalClassificationError = 0.53085938 * 2560; time = 0.1270s; samplesPerSecond = 20163.5
|
||||
MPI Rank 1: 08/16/2016 09:59:50: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95934862 * 2560; EvalClassificationError = 0.52812500 * 2560; time = 0.1253s; samplesPerSecond = 20436.0
|
||||
MPI Rank 1: 08/16/2016 09:59:50: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.94070839 * 2560; EvalClassificationError = 0.53125000 * 2560; time = 0.1236s; samplesPerSecond = 20712.5
|
||||
MPI Rank 1: 08/16/2016 09:59:50: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02105262 * 20480; EvalClassificationError = 0.54609375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.01478s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.92733488 * 83050; perplexity = 6.87117331; EvalErrorPrediction = 0.53122216 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.92733488 * 83050; EvalErrorPrediction = 0.53122216 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.92733488 * 83050; perplexity = 6.87117331; EvalClassificationError = 0.53122216 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.92733488 * 83050; EvalClassificationError = 0.53122216 * 83050
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94336419 * 10240; EvalErrorPrediction = 0.53056641 * 10240; time = 0.2793s; samplesPerSecond = 36660.2
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96525554 * 10240; EvalErrorPrediction = 0.54873047 * 10240; time = 0.2722s; samplesPerSecond = 37620.1
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95430987 * 20480; EvalErrorPrediction = 0.53964844 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.559871s
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94336419 * 10240; EvalClassificationError = 0.53056641 * 10240; time = 0.2793s; samplesPerSecond = 36660.2
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96525554 * 10240; EvalClassificationError = 0.54873047 * 10240; time = 0.2722s; samplesPerSecond = 37620.1
|
||||
MPI Rank 1: 08/16/2016 09:59:51: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95430987 * 20480; EvalClassificationError = 0.53964844 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.559871s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 09:59:52: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.90639119 * 83050; perplexity = 6.72876207; EvalErrorPrediction = 0.52304636 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:52: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.90639119 * 83050; EvalErrorPrediction = 0.52304636 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:52: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.90639119 * 83050; perplexity = 6.72876207; EvalClassificationError = 0.52304636 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:52: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.90639119 * 83050; EvalClassificationError = 0.52304636 * 83050
|
||||
MPI Rank 1: 08/16/2016 09:59:52: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:52: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -104,7 +104,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -130,7 +130,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -227,7 +227,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -253,7 +253,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -357,7 +357,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -383,7 +383,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -504,7 +504,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -533,7 +533,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -556,7 +556,7 @@ MPI Rank 0: 08/16/2016 03:01:59: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:01:59: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:01:59: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:01:59: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:01:59: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -611,73 +611,73 @@ MPI Rank 0: 08/16/2016 03:02:03: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.1102s; samplesPerSecond = 5809.0
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.1297s; samplesPerSecond = 4932.6
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.1064s; samplesPerSecond = 6015.2
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1118s; samplesPerSecond = 5724.0
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.1064s; samplesPerSecond = 6016.1
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.1079s; samplesPerSecond = 5932.8
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.1053s; samplesPerSecond = 6078.7
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.1123s; samplesPerSecond = 5700.3
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.1052s; samplesPerSecond = 6082.6
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.1119s; samplesPerSecond = 5721.2
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.1101s; samplesPerSecond = 5815.3
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.1088s; samplesPerSecond = 5884.1
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.1087s; samplesPerSecond = 5887.4
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.1054s; samplesPerSecond = 6071.6
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.1067s; samplesPerSecond = 5997.2
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.1076s; samplesPerSecond = 5947.6
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.1081s; samplesPerSecond = 5921.5
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.1094s; samplesPerSecond = 5849.2
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.1092s; samplesPerSecond = 5863.3
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.1069s; samplesPerSecond = 5984.8
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.1057s; samplesPerSecond = 6056.8
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.1077s; samplesPerSecond = 5940.4
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.1070s; samplesPerSecond = 5982.5
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1068s; samplesPerSecond = 5994.8
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.1059s; samplesPerSecond = 6041.3
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.1079s; samplesPerSecond = 5929.5
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.1071s; samplesPerSecond = 5975.4
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1073s; samplesPerSecond = 5964.1
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1080s; samplesPerSecond = 5926.3
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.1054s; samplesPerSecond = 6071.8
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.1064s; samplesPerSecond = 6016.1
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.1085s; samplesPerSecond = 5897.5
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.50129s
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.1102s; samplesPerSecond = 5809.0
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.1297s; samplesPerSecond = 4932.6
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.1064s; samplesPerSecond = 6015.2
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1118s; samplesPerSecond = 5724.0
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.1064s; samplesPerSecond = 6016.1
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.1079s; samplesPerSecond = 5932.8
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.1053s; samplesPerSecond = 6078.7
|
||||
MPI Rank 0: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.1123s; samplesPerSecond = 5700.3
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.1052s; samplesPerSecond = 6082.6
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.1119s; samplesPerSecond = 5721.2
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.1101s; samplesPerSecond = 5815.3
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.1088s; samplesPerSecond = 5884.1
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.1087s; samplesPerSecond = 5887.4
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.1054s; samplesPerSecond = 6071.6
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.1067s; samplesPerSecond = 5997.2
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.1076s; samplesPerSecond = 5947.6
|
||||
MPI Rank 0: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.1081s; samplesPerSecond = 5921.5
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.1094s; samplesPerSecond = 5849.2
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.1092s; samplesPerSecond = 5863.3
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.1069s; samplesPerSecond = 5984.8
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.1057s; samplesPerSecond = 6056.8
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.1077s; samplesPerSecond = 5940.4
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.1070s; samplesPerSecond = 5982.5
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1068s; samplesPerSecond = 5994.8
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.1059s; samplesPerSecond = 6041.3
|
||||
MPI Rank 0: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.1079s; samplesPerSecond = 5929.5
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.1071s; samplesPerSecond = 5975.4
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1073s; samplesPerSecond = 5964.1
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1080s; samplesPerSecond = 5926.3
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.1054s; samplesPerSecond = 6071.8
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.1064s; samplesPerSecond = 6016.1
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.1085s; samplesPerSecond = 5897.5
|
||||
MPI Rank 0: 08/16/2016 03:02:06: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.50129s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 10.0 ms
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.20991696 * 83050; perplexity = 9.11495943; EvalErrorPrediction = 0.60792294 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.20991696 * 83050; EvalErrorPrediction = 0.60792294 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.20991696 * 83050; perplexity = 9.11495943; EvalClassificationError = 0.60792294 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.20991696 * 83050; EvalClassificationError = 0.60792294 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:11: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelCrossValidation@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.09963072 * 2560; EvalErrorPrediction = 0.56132812 * 2560; time = 0.2349s; samplesPerSecond = 10898.6
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.02412622 * 2560; EvalErrorPrediction = 0.55000000 * 2560; time = 0.2179s; samplesPerSecond = 11749.9
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.00477550 * 2560; EvalErrorPrediction = 0.54296875 * 2560; time = 0.2211s; samplesPerSecond = 11580.5
|
||||
MPI Rank 0: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.99184610 * 2560; EvalErrorPrediction = 0.55273438 * 2560; time = 0.2159s; samplesPerSecond = 11858.8
|
||||
MPI Rank 0: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.98267472 * 2560; EvalErrorPrediction = 0.54023438 * 2560; time = 0.2193s; samplesPerSecond = 11675.5
|
||||
MPI Rank 0: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.93130832 * 2560; EvalErrorPrediction = 0.52773437 * 2560; time = 0.2181s; samplesPerSecond = 11735.8
|
||||
MPI Rank 0: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91975734 * 2560; EvalErrorPrediction = 0.51718750 * 2560; time = 0.2180s; samplesPerSecond = 11743.2
|
||||
MPI Rank 0: 08/16/2016 03:02:13: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.90691644 * 2560; EvalErrorPrediction = 0.52734375 * 2560; time = 0.2138s; samplesPerSecond = 11976.3
|
||||
MPI Rank 0: 08/16/2016 03:02:13: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.98262942 * 20480; EvalErrorPrediction = 0.53994141 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.77103s
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.09963072 * 2560; EvalClassificationError = 0.56132812 * 2560; time = 0.2349s; samplesPerSecond = 10898.6
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.02412622 * 2560; EvalClassificationError = 0.55000000 * 2560; time = 0.2179s; samplesPerSecond = 11749.9
|
||||
MPI Rank 0: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.00477550 * 2560; EvalClassificationError = 0.54296875 * 2560; time = 0.2211s; samplesPerSecond = 11580.5
|
||||
MPI Rank 0: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.99184610 * 2560; EvalClassificationError = 0.55273438 * 2560; time = 0.2159s; samplesPerSecond = 11858.8
|
||||
MPI Rank 0: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.98267472 * 2560; EvalClassificationError = 0.54023438 * 2560; time = 0.2193s; samplesPerSecond = 11675.5
|
||||
MPI Rank 0: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.93130832 * 2560; EvalClassificationError = 0.52773437 * 2560; time = 0.2181s; samplesPerSecond = 11735.8
|
||||
MPI Rank 0: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91975734 * 2560; EvalClassificationError = 0.51718750 * 2560; time = 0.2180s; samplesPerSecond = 11743.2
|
||||
MPI Rank 0: 08/16/2016 03:02:13: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.90691644 * 2560; EvalClassificationError = 0.52734375 * 2560; time = 0.2138s; samplesPerSecond = 11976.3
|
||||
MPI Rank 0: 08/16/2016 03:02:13: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.98262942 * 20480; EvalClassificationError = 0.53994141 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.77103s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: 08/16/2016 03:02:16: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.88974199 * 83050; perplexity = 6.61766107; EvalErrorPrediction = 0.52368453 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:16: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.88974199 * 83050; EvalErrorPrediction = 0.52368453 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:16: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.88974199 * 83050; perplexity = 6.61766107; EvalClassificationError = 0.52368453 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:16: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.88974199 * 83050; EvalClassificationError = 0.52368453 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:16: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelCrossValidation@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:16: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:16: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:02:17: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.90951347 * 10240; EvalErrorPrediction = 0.52617187 * 10240; time = 0.7135s; samplesPerSecond = 14352.0
|
||||
MPI Rank 0: 08/16/2016 03:02:17: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93082770 * 10240; EvalErrorPrediction = 0.54072266 * 10240; time = 0.6161s; samplesPerSecond = 16621.4
|
||||
MPI Rank 0: 08/16/2016 03:02:17: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.92017059 * 20480; EvalErrorPrediction = 0.53344727 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.3412s
|
||||
MPI Rank 0: 08/16/2016 03:02:17: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.90951347 * 10240; EvalClassificationError = 0.52617187 * 10240; time = 0.7135s; samplesPerSecond = 14352.0
|
||||
MPI Rank 0: 08/16/2016 03:02:17: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93082770 * 10240; EvalClassificationError = 0.54072266 * 10240; time = 0.6161s; samplesPerSecond = 16621.4
|
||||
MPI Rank 0: 08/16/2016 03:02:17: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.92017059 * 20480; EvalClassificationError = 0.53344727 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.3412s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: 08/16/2016 03:02:20: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.86850118 * 83050; perplexity = 6.47857889; EvalErrorPrediction = 0.51506321 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:20: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.86850118 * 83050; EvalErrorPrediction = 0.51506321 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:20: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.86850118 * 83050; perplexity = 6.47857889; EvalClassificationError = 0.51506321 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:20: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.86850118 * 83050; EvalClassificationError = 0.51506321 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:20: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelCrossValidation@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:02:20: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -731,7 +731,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -757,7 +757,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -854,7 +854,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -880,7 +880,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -984,7 +984,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1010,7 +1010,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1131,7 +1131,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1160,7 +1160,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1183,7 +1183,7 @@ MPI Rank 1: 08/16/2016 03:01:59: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:01:59: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:01:59: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:01:59: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:01:59: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1238,70 +1238,70 @@ MPI Rank 1: 08/16/2016 03:02:03: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.1103s; samplesPerSecond = 5804.6
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.1297s; samplesPerSecond = 4935.8
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.1064s; samplesPerSecond = 6015.3
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1118s; samplesPerSecond = 5724.9
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.1064s; samplesPerSecond = 6014.6
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.1078s; samplesPerSecond = 5935.0
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.1052s; samplesPerSecond = 6081.0
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.1123s; samplesPerSecond = 5698.7
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.1052s; samplesPerSecond = 6084.4
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.1120s; samplesPerSecond = 5715.3
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.1097s; samplesPerSecond = 5833.1
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.1090s; samplesPerSecond = 5873.0
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.1086s; samplesPerSecond = 5890.7
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.1054s; samplesPerSecond = 6072.0
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.1067s; samplesPerSecond = 5999.1
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.1075s; samplesPerSecond = 5953.5
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.1081s; samplesPerSecond = 5917.8
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.1095s; samplesPerSecond = 5845.5
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.1091s; samplesPerSecond = 5866.6
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.1069s; samplesPerSecond = 5988.2
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.1056s; samplesPerSecond = 6058.3
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.1077s; samplesPerSecond = 5942.6
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.1070s; samplesPerSecond = 5983.7
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1067s; samplesPerSecond = 5996.9
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.1061s; samplesPerSecond = 6033.9
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.1078s; samplesPerSecond = 5935.3
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.1071s; samplesPerSecond = 5976.9
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1072s; samplesPerSecond = 5967.7
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1081s; samplesPerSecond = 5922.4
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.1054s; samplesPerSecond = 6074.1
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.1063s; samplesPerSecond = 6019.5
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.1084s; samplesPerSecond = 5901.8
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.50129s
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.1103s; samplesPerSecond = 5804.6
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.1297s; samplesPerSecond = 4935.8
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.1064s; samplesPerSecond = 6015.3
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1118s; samplesPerSecond = 5724.9
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.1064s; samplesPerSecond = 6014.6
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.1078s; samplesPerSecond = 5935.0
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.1052s; samplesPerSecond = 6081.0
|
||||
MPI Rank 1: 08/16/2016 03:02:03: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.1123s; samplesPerSecond = 5698.7
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.1052s; samplesPerSecond = 6084.4
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.1120s; samplesPerSecond = 5715.3
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.1097s; samplesPerSecond = 5833.1
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.1090s; samplesPerSecond = 5873.0
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.1086s; samplesPerSecond = 5890.7
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.1054s; samplesPerSecond = 6072.0
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.1067s; samplesPerSecond = 5999.1
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.1075s; samplesPerSecond = 5953.5
|
||||
MPI Rank 1: 08/16/2016 03:02:04: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.1081s; samplesPerSecond = 5917.8
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.1095s; samplesPerSecond = 5845.5
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.1091s; samplesPerSecond = 5866.6
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.1069s; samplesPerSecond = 5988.2
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.1056s; samplesPerSecond = 6058.3
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.1077s; samplesPerSecond = 5942.6
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.1070s; samplesPerSecond = 5983.7
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1067s; samplesPerSecond = 5996.9
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.1061s; samplesPerSecond = 6033.9
|
||||
MPI Rank 1: 08/16/2016 03:02:05: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.1078s; samplesPerSecond = 5935.3
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.1071s; samplesPerSecond = 5976.9
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1072s; samplesPerSecond = 5967.7
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1081s; samplesPerSecond = 5922.4
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.1054s; samplesPerSecond = 6074.1
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.1063s; samplesPerSecond = 6019.5
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.1084s; samplesPerSecond = 5901.8
|
||||
MPI Rank 1: 08/16/2016 03:02:06: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.50129s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.20991696 * 83050; perplexity = 9.11495943; EvalErrorPrediction = 0.60792294 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.20991696 * 83050; EvalErrorPrediction = 0.60792294 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.20991696 * 83050; perplexity = 9.11495943; EvalClassificationError = 0.60792294 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.20991696 * 83050; EvalClassificationError = 0.60792294 * 83050
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.09963072 * 2560; EvalErrorPrediction = 0.56132812 * 2560; time = 0.2348s; samplesPerSecond = 10902.3
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.02412622 * 2560; EvalErrorPrediction = 0.55000000 * 2560; time = 0.2178s; samplesPerSecond = 11753.0
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.00477550 * 2560; EvalErrorPrediction = 0.54296875 * 2560; time = 0.2209s; samplesPerSecond = 11587.9
|
||||
MPI Rank 1: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.99184610 * 2560; EvalErrorPrediction = 0.55273438 * 2560; time = 0.2159s; samplesPerSecond = 11859.5
|
||||
MPI Rank 1: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.98267472 * 2560; EvalErrorPrediction = 0.54023438 * 2560; time = 0.2193s; samplesPerSecond = 11675.8
|
||||
MPI Rank 1: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.93130832 * 2560; EvalErrorPrediction = 0.52773437 * 2560; time = 0.2181s; samplesPerSecond = 11739.1
|
||||
MPI Rank 1: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91975734 * 2560; EvalErrorPrediction = 0.51718750 * 2560; time = 0.2179s; samplesPerSecond = 11747.4
|
||||
MPI Rank 1: 08/16/2016 03:02:13: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.90691644 * 2560; EvalErrorPrediction = 0.52734375 * 2560; time = 0.2137s; samplesPerSecond = 11980.4
|
||||
MPI Rank 1: 08/16/2016 03:02:13: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.98262942 * 20480; EvalErrorPrediction = 0.53994141 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.77102s
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.09963072 * 2560; EvalClassificationError = 0.56132812 * 2560; time = 0.2348s; samplesPerSecond = 10902.3
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.02412622 * 2560; EvalClassificationError = 0.55000000 * 2560; time = 0.2178s; samplesPerSecond = 11753.0
|
||||
MPI Rank 1: 08/16/2016 03:02:11: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.00477550 * 2560; EvalClassificationError = 0.54296875 * 2560; time = 0.2209s; samplesPerSecond = 11587.9
|
||||
MPI Rank 1: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.99184610 * 2560; EvalClassificationError = 0.55273438 * 2560; time = 0.2159s; samplesPerSecond = 11859.5
|
||||
MPI Rank 1: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.98267472 * 2560; EvalClassificationError = 0.54023438 * 2560; time = 0.2193s; samplesPerSecond = 11675.8
|
||||
MPI Rank 1: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.93130832 * 2560; EvalClassificationError = 0.52773437 * 2560; time = 0.2181s; samplesPerSecond = 11739.1
|
||||
MPI Rank 1: 08/16/2016 03:02:12: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91975734 * 2560; EvalClassificationError = 0.51718750 * 2560; time = 0.2179s; samplesPerSecond = 11747.4
|
||||
MPI Rank 1: 08/16/2016 03:02:13: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.90691644 * 2560; EvalClassificationError = 0.52734375 * 2560; time = 0.2137s; samplesPerSecond = 11980.4
|
||||
MPI Rank 1: 08/16/2016 03:02:13: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.98262942 * 20480; EvalClassificationError = 0.53994141 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.77102s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 03:02:16: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.88974199 * 83050; perplexity = 6.61766107; EvalErrorPrediction = 0.52368453 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:16: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.88974199 * 83050; EvalErrorPrediction = 0.52368453 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:16: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.88974199 * 83050; perplexity = 6.61766107; EvalClassificationError = 0.52368453 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:16: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.88974199 * 83050; EvalClassificationError = 0.52368453 * 83050
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:16: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:16: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:02:17: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.90951347 * 10240; EvalErrorPrediction = 0.52617187 * 10240; time = 0.7153s; samplesPerSecond = 14315.2
|
||||
MPI Rank 1: 08/16/2016 03:02:17: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93082770 * 10240; EvalErrorPrediction = 0.54072266 * 10240; time = 0.6162s; samplesPerSecond = 16618.6
|
||||
MPI Rank 1: 08/16/2016 03:02:17: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.92017059 * 20480; EvalErrorPrediction = 0.53344727 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.34119s
|
||||
MPI Rank 1: 08/16/2016 03:02:17: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.90951347 * 10240; EvalClassificationError = 0.52617187 * 10240; time = 0.7153s; samplesPerSecond = 14315.2
|
||||
MPI Rank 1: 08/16/2016 03:02:17: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93082770 * 10240; EvalClassificationError = 0.54072266 * 10240; time = 0.6162s; samplesPerSecond = 16618.6
|
||||
MPI Rank 1: 08/16/2016 03:02:17: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.92017059 * 20480; EvalClassificationError = 0.53344727 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=1.34119s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 03:02:20: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.86850118 * 83050; perplexity = 6.47857889; EvalErrorPrediction = 0.51506321 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:20: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.86850118 * 83050; EvalErrorPrediction = 0.51506321 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:20: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.86850118 * 83050; perplexity = 6.47857889; EvalClassificationError = 0.51506321 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:20: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.86850118 * 83050; EvalClassificationError = 0.51506321 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:20: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:20: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -104,7 +104,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -130,7 +130,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -227,7 +227,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -253,7 +253,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -357,7 +357,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -383,7 +383,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -505,7 +505,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -534,7 +534,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -557,7 +557,7 @@ MPI Rank 0: 08/16/2016 03:02:30: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:02:30: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:30: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:02:30: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:02:30: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -612,73 +612,73 @@ MPI Rank 0: 08/16/2016 03:02:36: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1600s; samplesPerSecond = 3999.8
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.1186s; samplesPerSecond = 5394.3
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.1100s; samplesPerSecond = 5815.7
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1132s; samplesPerSecond = 5654.1
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.1101s; samplesPerSecond = 5814.3
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.1099s; samplesPerSecond = 5821.8
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.1099s; samplesPerSecond = 5822.4
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.1093s; samplesPerSecond = 5852.9
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.1090s; samplesPerSecond = 5869.5
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.1094s; samplesPerSecond = 5848.0
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.1095s; samplesPerSecond = 5843.8
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.1112s; samplesPerSecond = 5754.0
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.1094s; samplesPerSecond = 5851.4
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.1098s; samplesPerSecond = 5826.3
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.1093s; samplesPerSecond = 5855.0
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.1103s; samplesPerSecond = 5801.9
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.1119s; samplesPerSecond = 5720.9
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1099s; samplesPerSecond = 5822.0
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.1097s; samplesPerSecond = 5832.3
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.1112s; samplesPerSecond = 5755.8
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1138s; samplesPerSecond = 5624.2
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1096s; samplesPerSecond = 5840.1
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1095s; samplesPerSecond = 5843.5
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1092s; samplesPerSecond = 5858.4
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.1097s; samplesPerSecond = 5836.1
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.1097s; samplesPerSecond = 5835.8
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1101s; samplesPerSecond = 5811.8
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.1094s; samplesPerSecond = 5848.0
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.1098s; samplesPerSecond = 5830.1
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.1096s; samplesPerSecond = 5837.3
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.1098s; samplesPerSecond = 5827.2
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.1109s; samplesPerSecond = 5771.4
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.60518s
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1600s; samplesPerSecond = 3999.8
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.1186s; samplesPerSecond = 5394.3
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.1100s; samplesPerSecond = 5815.7
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1132s; samplesPerSecond = 5654.1
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.1101s; samplesPerSecond = 5814.3
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.1099s; samplesPerSecond = 5821.8
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.1099s; samplesPerSecond = 5822.4
|
||||
MPI Rank 0: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.1093s; samplesPerSecond = 5852.9
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.1090s; samplesPerSecond = 5869.5
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.1094s; samplesPerSecond = 5848.0
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.1095s; samplesPerSecond = 5843.8
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.1112s; samplesPerSecond = 5754.0
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.1094s; samplesPerSecond = 5851.4
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.1098s; samplesPerSecond = 5826.3
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.1093s; samplesPerSecond = 5855.0
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.1103s; samplesPerSecond = 5801.9
|
||||
MPI Rank 0: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.1119s; samplesPerSecond = 5720.9
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1099s; samplesPerSecond = 5822.0
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.1097s; samplesPerSecond = 5832.3
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.1112s; samplesPerSecond = 5755.8
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1138s; samplesPerSecond = 5624.2
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1096s; samplesPerSecond = 5840.1
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1095s; samplesPerSecond = 5843.5
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1092s; samplesPerSecond = 5858.4
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.1097s; samplesPerSecond = 5836.1
|
||||
MPI Rank 0: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.1097s; samplesPerSecond = 5835.8
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1101s; samplesPerSecond = 5811.8
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.1094s; samplesPerSecond = 5848.0
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.1098s; samplesPerSecond = 5830.1
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.1096s; samplesPerSecond = 5837.3
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.1098s; samplesPerSecond = 5827.2
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.1109s; samplesPerSecond = 5771.4
|
||||
MPI Rank 0: 08/16/2016 03:02:39: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.60518s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: requiredata: determined feature kind as 33-dimensional 'USER' with frame shift 10.0 ms
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.24821047 * 83050; perplexity = 9.47077247; EvalErrorPrediction = 0.61623119 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.24821047 * 83050; EvalErrorPrediction = 0.61623119 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.24821047 * 83050; perplexity = 9.47077247; EvalClassificationError = 0.61623119 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.24821047 * 83050; EvalClassificationError = 0.61623119 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:43: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelCrossValidation@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.13894071 * 2560; EvalErrorPrediction = 0.56992188 * 2560; time = 0.1674s; samplesPerSecond = 15289.7
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.06106261 * 2560; EvalErrorPrediction = 0.55664063 * 2560; time = 0.1479s; samplesPerSecond = 17309.5
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.04459475 * 2560; EvalErrorPrediction = 0.55039063 * 2560; time = 0.1413s; samplesPerSecond = 18120.8
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.03347291 * 2560; EvalErrorPrediction = 0.55742187 * 2560; time = 0.1455s; samplesPerSecond = 17598.9
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02079287 * 2560; EvalErrorPrediction = 0.54414063 * 2560; time = 0.1409s; samplesPerSecond = 18169.6
|
||||
MPI Rank 0: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96950012 * 2560; EvalErrorPrediction = 0.53085938 * 2560; time = 0.1470s; samplesPerSecond = 17416.0
|
||||
MPI Rank 0: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95934862 * 2560; EvalErrorPrediction = 0.52812500 * 2560; time = 0.1403s; samplesPerSecond = 18247.5
|
||||
MPI Rank 0: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.94070839 * 2560; EvalErrorPrediction = 0.53125000 * 2560; time = 0.1403s; samplesPerSecond = 18240.4
|
||||
MPI Rank 0: 08/16/2016 03:02:44: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02105262 * 20480; EvalErrorPrediction = 0.54609375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.1819s
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.13894071 * 2560; EvalClassificationError = 0.56992188 * 2560; time = 0.1674s; samplesPerSecond = 15289.7
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.06106261 * 2560; EvalClassificationError = 0.55664063 * 2560; time = 0.1479s; samplesPerSecond = 17309.5
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.04459475 * 2560; EvalClassificationError = 0.55039063 * 2560; time = 0.1413s; samplesPerSecond = 18120.8
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.03347291 * 2560; EvalClassificationError = 0.55742187 * 2560; time = 0.1455s; samplesPerSecond = 17598.9
|
||||
MPI Rank 0: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02079287 * 2560; EvalClassificationError = 0.54414063 * 2560; time = 0.1409s; samplesPerSecond = 18169.6
|
||||
MPI Rank 0: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96950012 * 2560; EvalClassificationError = 0.53085938 * 2560; time = 0.1470s; samplesPerSecond = 17416.0
|
||||
MPI Rank 0: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95934862 * 2560; EvalClassificationError = 0.52812500 * 2560; time = 0.1403s; samplesPerSecond = 18247.5
|
||||
MPI Rank 0: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.94070839 * 2560; EvalClassificationError = 0.53125000 * 2560; time = 0.1403s; samplesPerSecond = 18240.4
|
||||
MPI Rank 0: 08/16/2016 03:02:44: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02105262 * 20480; EvalClassificationError = 0.54609375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.1819s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: 08/16/2016 03:02:45: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.92733488 * 83050; perplexity = 6.87117331; EvalErrorPrediction = 0.53122216 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:45: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.92733488 * 83050; EvalErrorPrediction = 0.53122216 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:45: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.92733488 * 83050; perplexity = 6.87117331; EvalClassificationError = 0.53122216 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:45: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.92733488 * 83050; EvalClassificationError = 0.53122216 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:45: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelCrossValidation@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:45: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:45: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:02:46: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94336419 * 10240; EvalErrorPrediction = 0.53056641 * 10240; time = 0.3479s; samplesPerSecond = 29432.6
|
||||
MPI Rank 0: 08/16/2016 03:02:46: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96525554 * 10240; EvalErrorPrediction = 0.54873047 * 10240; time = 0.3252s; samplesPerSecond = 31484.1
|
||||
MPI Rank 0: 08/16/2016 03:02:46: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95430987 * 20480; EvalErrorPrediction = 0.53964844 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.684367s
|
||||
MPI Rank 0: 08/16/2016 03:02:46: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94336419 * 10240; EvalClassificationError = 0.53056641 * 10240; time = 0.3479s; samplesPerSecond = 29432.6
|
||||
MPI Rank 0: 08/16/2016 03:02:46: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96525554 * 10240; EvalClassificationError = 0.54873047 * 10240; time = 0.3252s; samplesPerSecond = 31484.1
|
||||
MPI Rank 0: 08/16/2016 03:02:46: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95430987 * 20480; EvalClassificationError = 0.53964844 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.684367s
|
||||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 0 of 2, with 1 datapasses
|
||||
MPI Rank 0: 08/16/2016 03:02:47: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.90639119 * 83050; perplexity = 6.72876207; EvalErrorPrediction = 0.52304636 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:47: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.90639119 * 83050; EvalErrorPrediction = 0.52304636 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:47: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.90639119 * 83050; perplexity = 6.72876207; EvalClassificationError = 0.52304636 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:47: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.90639119 * 83050; EvalClassificationError = 0.52304636 * 83050
|
||||
MPI Rank 0: 08/16/2016 03:02:47: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelCrossValidation@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:02:47: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -732,7 +732,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -758,7 +758,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -855,7 +855,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -881,7 +881,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -985,7 +985,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1011,7 +1011,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='eval')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='eval')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1133,7 +1133,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1162,7 +1162,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1185,7 +1185,7 @@ MPI Rank 1: 08/16/2016 03:02:31: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:02:31: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:31: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:02:31: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:02:31: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1240,70 +1240,70 @@ MPI Rank 1: 08/16/2016 03:02:36: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1549s; samplesPerSecond = 4132.8
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.1179s; samplesPerSecond = 5427.5
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.1100s; samplesPerSecond = 5817.1
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1131s; samplesPerSecond = 5656.4
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.1101s; samplesPerSecond = 5814.9
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.1099s; samplesPerSecond = 5822.0
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.1099s; samplesPerSecond = 5822.3
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.1093s; samplesPerSecond = 5855.2
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.1090s; samplesPerSecond = 5869.0
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.1094s; samplesPerSecond = 5847.7
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.1095s; samplesPerSecond = 5845.2
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.1120s; samplesPerSecond = 5714.6
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.1094s; samplesPerSecond = 5851.4
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.1099s; samplesPerSecond = 5826.1
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.1093s; samplesPerSecond = 5856.2
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.1103s; samplesPerSecond = 5802.0
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.1111s; samplesPerSecond = 5758.2
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1099s; samplesPerSecond = 5823.7
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.1097s; samplesPerSecond = 5832.0
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.1125s; samplesPerSecond = 5691.4
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1119s; samplesPerSecond = 5717.3
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1096s; samplesPerSecond = 5839.5
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1095s; samplesPerSecond = 5844.2
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.1092s; samplesPerSecond = 5860.4
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.1097s; samplesPerSecond = 5836.5
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.1096s; samplesPerSecond = 5837.6
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1101s; samplesPerSecond = 5811.7
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.1094s; samplesPerSecond = 5850.6
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.1098s; samplesPerSecond = 5829.5
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.1096s; samplesPerSecond = 5839.4
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.1098s; samplesPerSecond = 5826.7
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.1109s; samplesPerSecond = 5772.2
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.60483s
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1549s; samplesPerSecond = 4132.8
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.1179s; samplesPerSecond = 5427.5
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.1100s; samplesPerSecond = 5817.1
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1131s; samplesPerSecond = 5656.4
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.1101s; samplesPerSecond = 5814.9
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.1099s; samplesPerSecond = 5822.0
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.1099s; samplesPerSecond = 5822.3
|
||||
MPI Rank 1: 08/16/2016 03:02:36: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.1093s; samplesPerSecond = 5855.2
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.1090s; samplesPerSecond = 5869.0
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.1094s; samplesPerSecond = 5847.7
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.1095s; samplesPerSecond = 5845.2
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.1120s; samplesPerSecond = 5714.6
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.1094s; samplesPerSecond = 5851.4
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.1099s; samplesPerSecond = 5826.1
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.1093s; samplesPerSecond = 5856.2
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.1103s; samplesPerSecond = 5802.0
|
||||
MPI Rank 1: 08/16/2016 03:02:37: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.1111s; samplesPerSecond = 5758.2
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1099s; samplesPerSecond = 5823.7
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.1097s; samplesPerSecond = 5832.0
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.1125s; samplesPerSecond = 5691.4
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1119s; samplesPerSecond = 5717.3
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1096s; samplesPerSecond = 5839.5
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1095s; samplesPerSecond = 5844.2
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.1092s; samplesPerSecond = 5860.4
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.1097s; samplesPerSecond = 5836.5
|
||||
MPI Rank 1: 08/16/2016 03:02:38: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.1096s; samplesPerSecond = 5837.6
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1101s; samplesPerSecond = 5811.7
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.1094s; samplesPerSecond = 5850.6
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.1098s; samplesPerSecond = 5829.5
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.1096s; samplesPerSecond = 5839.4
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.1098s; samplesPerSecond = 5826.7
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.1109s; samplesPerSecond = 5772.2
|
||||
MPI Rank 1: 08/16/2016 03:02:39: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.60483s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.24821047 * 83050; perplexity = 9.47077247; EvalErrorPrediction = 0.61623119 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.24821047 * 83050; EvalErrorPrediction = 0.61623119 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Final Results: Minibatch[1-1299]: CrossEntropyWithSoftmax = 2.24821047 * 83050; perplexity = 9.47077247; EvalClassificationError = 0.61623119 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Finished Epoch[ 1 of 3]: [Validate] CrossEntropyWithSoftmax = 2.24821047 * 83050; EvalClassificationError = 0.61623119 * 83050
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.13894071 * 2560; EvalErrorPrediction = 0.56992188 * 2560; time = 0.1664s; samplesPerSecond = 15385.0
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.06106261 * 2560; EvalErrorPrediction = 0.55664063 * 2560; time = 0.1471s; samplesPerSecond = 17399.5
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.04459475 * 2560; EvalErrorPrediction = 0.55039063 * 2560; time = 0.1413s; samplesPerSecond = 18123.8
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.03347291 * 2560; EvalErrorPrediction = 0.55742187 * 2560; time = 0.1462s; samplesPerSecond = 17508.3
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02079287 * 2560; EvalErrorPrediction = 0.54414063 * 2560; time = 0.1409s; samplesPerSecond = 18169.7
|
||||
MPI Rank 1: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96950012 * 2560; EvalErrorPrediction = 0.53085938 * 2560; time = 0.1462s; samplesPerSecond = 17507.1
|
||||
MPI Rank 1: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95934862 * 2560; EvalErrorPrediction = 0.52812500 * 2560; time = 0.1410s; samplesPerSecond = 18150.4
|
||||
MPI Rank 1: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.94070839 * 2560; EvalErrorPrediction = 0.53125000 * 2560; time = 0.1403s; samplesPerSecond = 18241.9
|
||||
MPI Rank 1: 08/16/2016 03:02:44: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02105262 * 20480; EvalErrorPrediction = 0.54609375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.18217s
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.13894071 * 2560; EvalClassificationError = 0.56992188 * 2560; time = 0.1664s; samplesPerSecond = 15385.0
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.06106261 * 2560; EvalClassificationError = 0.55664063 * 2560; time = 0.1471s; samplesPerSecond = 17399.5
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.04459475 * 2560; EvalClassificationError = 0.55039063 * 2560; time = 0.1413s; samplesPerSecond = 18123.8
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.03347291 * 2560; EvalClassificationError = 0.55742187 * 2560; time = 0.1462s; samplesPerSecond = 17508.3
|
||||
MPI Rank 1: 08/16/2016 03:02:43: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.02079287 * 2560; EvalClassificationError = 0.54414063 * 2560; time = 0.1409s; samplesPerSecond = 18169.7
|
||||
MPI Rank 1: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.96950012 * 2560; EvalClassificationError = 0.53085938 * 2560; time = 0.1462s; samplesPerSecond = 17507.1
|
||||
MPI Rank 1: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.95934862 * 2560; EvalClassificationError = 0.52812500 * 2560; time = 0.1410s; samplesPerSecond = 18150.4
|
||||
MPI Rank 1: 08/16/2016 03:02:44: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.94070839 * 2560; EvalClassificationError = 0.53125000 * 2560; time = 0.1403s; samplesPerSecond = 18241.9
|
||||
MPI Rank 1: 08/16/2016 03:02:44: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.02105262 * 20480; EvalClassificationError = 0.54609375 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.18217s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 03:02:45: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.92733488 * 83050; perplexity = 6.87117331; EvalErrorPrediction = 0.53122216 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:45: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.92733488 * 83050; EvalErrorPrediction = 0.53122216 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:45: Final Results: Minibatch[1-326]: CrossEntropyWithSoftmax = 1.92733488 * 83050; perplexity = 6.87117331; EvalClassificationError = 0.53122216 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:45: Finished Epoch[ 2 of 3]: [Validate] CrossEntropyWithSoftmax = 1.92733488 * 83050; EvalClassificationError = 0.53122216 * 83050
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:45: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:45: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 2, NumGradientBits = 64), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:02:46: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94336419 * 10240; EvalErrorPrediction = 0.53056641 * 10240; time = 0.3456s; samplesPerSecond = 29626.5
|
||||
MPI Rank 1: 08/16/2016 03:02:46: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96525554 * 10240; EvalErrorPrediction = 0.54873047 * 10240; time = 0.3253s; samplesPerSecond = 31482.2
|
||||
MPI Rank 1: 08/16/2016 03:02:46: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95430987 * 20480; EvalErrorPrediction = 0.53964844 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.684319s
|
||||
MPI Rank 1: 08/16/2016 03:02:46: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94336419 * 10240; EvalClassificationError = 0.53056641 * 10240; time = 0.3456s; samplesPerSecond = 29626.5
|
||||
MPI Rank 1: 08/16/2016 03:02:46: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.96525554 * 10240; EvalClassificationError = 0.54873047 * 10240; time = 0.3253s; samplesPerSecond = 31482.2
|
||||
MPI Rank 1: 08/16/2016 03:02:46: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.95430987 * 20480; EvalClassificationError = 0.53964844 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.684319s
|
||||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..83050] (first utterance at frame 0), data subset 1 of 2, with 1 datapasses
|
||||
MPI Rank 1: 08/16/2016 03:02:47: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.90639119 * 83050; perplexity = 6.72876207; EvalErrorPrediction = 0.52304636 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:47: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.90639119 * 83050; EvalErrorPrediction = 0.52304636 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:47: Final Results: Minibatch[1-83]: CrossEntropyWithSoftmax = 1.90639119 * 83050; perplexity = 6.72876207; EvalClassificationError = 0.52304636 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:47: Finished Epoch[ 3 of 3]: [Validate] CrossEntropyWithSoftmax = 1.90639119 * 83050; EvalClassificationError = 0.52304636 * 83050
|
||||
MPI Rank 1: 08/16/2016 03:02:47: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:47: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -133,7 +133,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -159,7 +159,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -236,7 +236,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -262,7 +262,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -345,7 +345,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -371,7 +371,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -467,7 +467,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -496,7 +496,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -519,7 +519,7 @@ MPI Rank 0: 08/16/2016 09:59:54: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 09:59:54: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:54: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 09:59:54: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 09:59:54: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -572,63 +572,63 @@ MPI Rank 0: 08/16/2016 09:59:56: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:56: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62304965 * 640; EvalErrorPrediction = 0.93437500 * 640; time = 0.2660s; samplesPerSecond = 2405.9
|
||||
MPI Rank 0: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.38132581 * 640; EvalErrorPrediction = 0.93125000 * 640; time = 0.1966s; samplesPerSecond = 3255.6
|
||||
MPI Rank 0: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98762394 * 640; EvalErrorPrediction = 0.88593750 * 640; time = 0.4635s; samplesPerSecond = 1380.7
|
||||
MPI Rank 0: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74489685 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1425s; samplesPerSecond = 4491.0
|
||||
MPI Rank 0: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83505115 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.3191s; samplesPerSecond = 2005.3
|
||||
MPI Rank 0: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71847012 * 640; EvalErrorPrediction = 0.87187500 * 640; time = 0.1449s; samplesPerSecond = 4416.4
|
||||
MPI Rank 0: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.43056226 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.1763s; samplesPerSecond = 3631.1
|
||||
MPI Rank 0: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.55346588 * 640; EvalErrorPrediction = 0.82656250 * 640; time = 0.3509s; samplesPerSecond = 1823.9
|
||||
MPI Rank 0: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.52102692 * 640; EvalErrorPrediction = 0.82343750 * 640; time = 0.1122s; samplesPerSecond = 5706.3
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.43553680 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.1295s; samplesPerSecond = 4943.5
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.53125248 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.1747s; samplesPerSecond = 3663.6
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.30802761 * 640; EvalErrorPrediction = 0.80000000 * 640; time = 0.0991s; samplesPerSecond = 6458.3
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.23431326 * 640; EvalErrorPrediction = 0.81562500 * 640; time = 0.1343s; samplesPerSecond = 4766.2
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.09809151 * 640; EvalErrorPrediction = 0.75312500 * 640; time = 0.1553s; samplesPerSecond = 4121.9
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.99690035 * 640; EvalErrorPrediction = 0.74843750 * 640; time = 0.3007s; samplesPerSecond = 2128.5
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.14211613 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.1090s; samplesPerSecond = 5869.2
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.86222858 * 640; EvalErrorPrediction = 0.72343750 * 640; time = 0.1125s; samplesPerSecond = 5689.9
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.75962647 * 640; EvalErrorPrediction = 0.65937500 * 640; time = 0.1029s; samplesPerSecond = 6218.7
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.84542338 * 640; EvalErrorPrediction = 0.72031250 * 640; time = 0.0811s; samplesPerSecond = 7893.3
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.75031878 * 640; EvalErrorPrediction = 0.68281250 * 640; time = 0.1661s; samplesPerSecond = 3853.3
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.62936500 * 640; EvalErrorPrediction = 0.66562500 * 640; time = 0.1480s; samplesPerSecond = 4323.5
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.62926388 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.2952s; samplesPerSecond = 2168.0
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.55365661 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.1730s; samplesPerSecond = 3698.5
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.58641615 * 640; EvalErrorPrediction = 0.66250000 * 640; time = 0.1207s; samplesPerSecond = 5301.5
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50844204 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.1571s; samplesPerSecond = 4073.7
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.42228794 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.1795s; samplesPerSecond = 3565.7
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.28812796 * 640; EvalErrorPrediction = 0.59375000 * 640; time = 0.1826s; samplesPerSecond = 3505.4
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.54392000 * 640; EvalErrorPrediction = 0.66718750 * 640; time = 0.1485s; samplesPerSecond = 4310.4
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.29038551 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.3335s; samplesPerSecond = 1918.9
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.27576953 * 640; EvalErrorPrediction = 0.61406250 * 640; time = 0.0984s; samplesPerSecond = 6506.5
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.26696531 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1104s; samplesPerSecond = 5796.3
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.26576222 * 640; EvalErrorPrediction = 0.61093750 * 640; time = 0.1931s; samplesPerSecond = 3314.3
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.06311469 * 20480; EvalErrorPrediction = 0.73945313 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.88542s
|
||||
MPI Rank 0: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62304965 * 640; EvalClassificationError = 0.93437500 * 640; time = 0.2660s; samplesPerSecond = 2405.9
|
||||
MPI Rank 0: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.38132581 * 640; EvalClassificationError = 0.93125000 * 640; time = 0.1966s; samplesPerSecond = 3255.6
|
||||
MPI Rank 0: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98762394 * 640; EvalClassificationError = 0.88593750 * 640; time = 0.4635s; samplesPerSecond = 1380.7
|
||||
MPI Rank 0: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74489685 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1425s; samplesPerSecond = 4491.0
|
||||
MPI Rank 0: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83505115 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.3191s; samplesPerSecond = 2005.3
|
||||
MPI Rank 0: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71847012 * 640; EvalClassificationError = 0.87187500 * 640; time = 0.1449s; samplesPerSecond = 4416.4
|
||||
MPI Rank 0: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.43056226 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.1763s; samplesPerSecond = 3631.1
|
||||
MPI Rank 0: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.55346588 * 640; EvalClassificationError = 0.82656250 * 640; time = 0.3509s; samplesPerSecond = 1823.9
|
||||
MPI Rank 0: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.52102692 * 640; EvalClassificationError = 0.82343750 * 640; time = 0.1122s; samplesPerSecond = 5706.3
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.43553680 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.1295s; samplesPerSecond = 4943.5
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.53125248 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.1747s; samplesPerSecond = 3663.6
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.30802761 * 640; EvalClassificationError = 0.80000000 * 640; time = 0.0991s; samplesPerSecond = 6458.3
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.23431326 * 640; EvalClassificationError = 0.81562500 * 640; time = 0.1343s; samplesPerSecond = 4766.2
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.09809151 * 640; EvalClassificationError = 0.75312500 * 640; time = 0.1553s; samplesPerSecond = 4121.9
|
||||
MPI Rank 0: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.99690035 * 640; EvalClassificationError = 0.74843750 * 640; time = 0.3007s; samplesPerSecond = 2128.5
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.14211613 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.1090s; samplesPerSecond = 5869.2
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.86222858 * 640; EvalClassificationError = 0.72343750 * 640; time = 0.1125s; samplesPerSecond = 5689.9
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.75962647 * 640; EvalClassificationError = 0.65937500 * 640; time = 0.1029s; samplesPerSecond = 6218.7
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.84542338 * 640; EvalClassificationError = 0.72031250 * 640; time = 0.0811s; samplesPerSecond = 7893.3
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.75031878 * 640; EvalClassificationError = 0.68281250 * 640; time = 0.1661s; samplesPerSecond = 3853.3
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.62936500 * 640; EvalClassificationError = 0.66562500 * 640; time = 0.1480s; samplesPerSecond = 4323.5
|
||||
MPI Rank 0: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.62926388 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.2952s; samplesPerSecond = 2168.0
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.55365661 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.1730s; samplesPerSecond = 3698.5
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.58641615 * 640; EvalClassificationError = 0.66250000 * 640; time = 0.1207s; samplesPerSecond = 5301.5
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50844204 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.1571s; samplesPerSecond = 4073.7
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.42228794 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.1795s; samplesPerSecond = 3565.7
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.28812796 * 640; EvalClassificationError = 0.59375000 * 640; time = 0.1826s; samplesPerSecond = 3505.4
|
||||
MPI Rank 0: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.54392000 * 640; EvalClassificationError = 0.66718750 * 640; time = 0.1485s; samplesPerSecond = 4310.4
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.29038551 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.3335s; samplesPerSecond = 1918.9
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.27576953 * 640; EvalClassificationError = 0.61406250 * 640; time = 0.0984s; samplesPerSecond = 6506.5
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.26696531 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1104s; samplesPerSecond = 5796.3
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.26576222 * 640; EvalClassificationError = 0.61093750 * 640; time = 0.1931s; samplesPerSecond = 3314.3
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.06311469 * 20480; EvalClassificationError = 0.73945313 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.88542s
|
||||
MPI Rank 0: 08/16/2016 10:00:02: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelNoQuantization@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:00:02: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.16726831 * 2560; EvalErrorPrediction = 0.57851562 * 2560; time = 0.5967s; samplesPerSecond = 4290.2
|
||||
MPI Rank 0: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.08752017 * 2560; EvalErrorPrediction = 0.56523437 * 2560; time = 0.3821s; samplesPerSecond = 6699.7
|
||||
MPI Rank 0: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07134545 * 2560; EvalErrorPrediction = 0.55585938 * 2560; time = 0.6943s; samplesPerSecond = 3686.9
|
||||
MPI Rank 0: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.05824888 * 2560; EvalErrorPrediction = 0.56562500 * 2560; time = 0.3895s; samplesPerSecond = 6572.7
|
||||
MPI Rank 0: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04811411 * 2560; EvalErrorPrediction = 0.55937500 * 2560; time = 0.2381s; samplesPerSecond = 10753.5
|
||||
MPI Rank 0: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.99263458 * 2560; EvalErrorPrediction = 0.54375000 * 2560; time = 0.5332s; samplesPerSecond = 4801.1
|
||||
MPI Rank 0: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99005701 * 2560; EvalErrorPrediction = 0.53359375 * 2560; time = 0.4091s; samplesPerSecond = 6257.6
|
||||
MPI Rank 0: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.97189564 * 2560; EvalErrorPrediction = 0.53984375 * 2560; time = 0.2384s; samplesPerSecond = 10739.2
|
||||
MPI Rank 0: 08/16/2016 10:00:06: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04838552 * 20480; EvalErrorPrediction = 0.55522461 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=3.48718s
|
||||
MPI Rank 0: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.16726831 * 2560; EvalClassificationError = 0.57851562 * 2560; time = 0.5967s; samplesPerSecond = 4290.2
|
||||
MPI Rank 0: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.08752017 * 2560; EvalClassificationError = 0.56523437 * 2560; time = 0.3821s; samplesPerSecond = 6699.7
|
||||
MPI Rank 0: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07134545 * 2560; EvalClassificationError = 0.55585938 * 2560; time = 0.6943s; samplesPerSecond = 3686.9
|
||||
MPI Rank 0: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.05824888 * 2560; EvalClassificationError = 0.56562500 * 2560; time = 0.3895s; samplesPerSecond = 6572.7
|
||||
MPI Rank 0: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04811411 * 2560; EvalClassificationError = 0.55937500 * 2560; time = 0.2381s; samplesPerSecond = 10753.5
|
||||
MPI Rank 0: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.99263458 * 2560; EvalClassificationError = 0.54375000 * 2560; time = 0.5332s; samplesPerSecond = 4801.1
|
||||
MPI Rank 0: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99005701 * 2560; EvalClassificationError = 0.53359375 * 2560; time = 0.4091s; samplesPerSecond = 6257.6
|
||||
MPI Rank 0: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.97189564 * 2560; EvalClassificationError = 0.53984375 * 2560; time = 0.2384s; samplesPerSecond = 10739.2
|
||||
MPI Rank 0: 08/16/2016 10:00:06: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04838552 * 20480; EvalClassificationError = 0.55522461 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=3.48718s
|
||||
MPI Rank 0: 08/16/2016 10:00:06: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelNoQuantization@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:00:06: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:00:06: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:00:07: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97114275 * 10240; EvalErrorPrediction = 0.53769531 * 10240; time = 0.9884s; samplesPerSecond = 10360.6
|
||||
MPI Rank 0: 08/16/2016 10:00:08: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.99390313 * 10240; EvalErrorPrediction = 0.55712891 * 10240; time = 1.4042s; samplesPerSecond = 7292.7
|
||||
MPI Rank 0: 08/16/2016 10:00:08: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.98252294 * 20480; EvalErrorPrediction = 0.54741211 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.39916s
|
||||
MPI Rank 0: 08/16/2016 10:00:07: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97114275 * 10240; EvalClassificationError = 0.53769531 * 10240; time = 0.9884s; samplesPerSecond = 10360.6
|
||||
MPI Rank 0: 08/16/2016 10:00:08: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.99390313 * 10240; EvalClassificationError = 0.55712891 * 10240; time = 1.4042s; samplesPerSecond = 7292.7
|
||||
MPI Rank 0: 08/16/2016 10:00:08: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.98252294 * 20480; EvalClassificationError = 0.54741211 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.39916s
|
||||
MPI Rank 0: 08/16/2016 10:00:08: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelNoQuantization@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 10:00:08: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -681,7 +681,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -707,7 +707,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -784,7 +784,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -810,7 +810,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -893,7 +893,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -919,7 +919,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1015,7 +1015,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1044,7 +1044,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1067,7 +1067,7 @@ MPI Rank 1: 08/16/2016 09:59:54: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 09:59:54: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:54: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 09:59:54: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 09:59:54: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1120,61 +1120,61 @@ MPI Rank 1: 08/16/2016 09:59:56: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:56: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62304965 * 640; EvalErrorPrediction = 0.93437500 * 640; time = 0.2613s; samplesPerSecond = 2449.3
|
||||
MPI Rank 1: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.38132581 * 640; EvalErrorPrediction = 0.93125000 * 640; time = 0.1965s; samplesPerSecond = 3256.6
|
||||
MPI Rank 1: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98762394 * 640; EvalErrorPrediction = 0.88593750 * 640; time = 0.4674s; samplesPerSecond = 1369.3
|
||||
MPI Rank 1: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74489685 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1429s; samplesPerSecond = 4478.8
|
||||
MPI Rank 1: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83505115 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.3190s; samplesPerSecond = 2006.1
|
||||
MPI Rank 1: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71847012 * 640; EvalErrorPrediction = 0.87187500 * 640; time = 0.1405s; samplesPerSecond = 4554.7
|
||||
MPI Rank 1: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.43056226 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.1769s; samplesPerSecond = 3617.6
|
||||
MPI Rank 1: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.55346588 * 640; EvalErrorPrediction = 0.82656250 * 640; time = 0.3548s; samplesPerSecond = 1803.8
|
||||
MPI Rank 1: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.52102692 * 640; EvalErrorPrediction = 0.82343750 * 640; time = 0.1120s; samplesPerSecond = 5711.7
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.43553680 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.1256s; samplesPerSecond = 5094.0
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.53125248 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.1788s; samplesPerSecond = 3579.7
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.30802761 * 640; EvalErrorPrediction = 0.80000000 * 640; time = 0.0993s; samplesPerSecond = 6447.6
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.23431326 * 640; EvalErrorPrediction = 0.81562500 * 640; time = 0.1337s; samplesPerSecond = 4787.0
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.09809151 * 640; EvalErrorPrediction = 0.75312500 * 640; time = 0.1556s; samplesPerSecond = 4112.8
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.99690035 * 640; EvalErrorPrediction = 0.74843750 * 640; time = 0.3005s; samplesPerSecond = 2129.6
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.14211613 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.1092s; samplesPerSecond = 5859.1
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.86222858 * 640; EvalErrorPrediction = 0.72343750 * 640; time = 0.1125s; samplesPerSecond = 5688.0
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.75962647 * 640; EvalErrorPrediction = 0.65937500 * 640; time = 0.1028s; samplesPerSecond = 6225.1
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.84542338 * 640; EvalErrorPrediction = 0.72031250 * 640; time = 0.0814s; samplesPerSecond = 7864.6
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.75031878 * 640; EvalErrorPrediction = 0.68281250 * 640; time = 0.1590s; samplesPerSecond = 4025.9
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.62936500 * 640; EvalErrorPrediction = 0.66562500 * 640; time = 0.1509s; samplesPerSecond = 4240.3
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.62926388 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.1572s; samplesPerSecond = 4071.4
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.55365661 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.3110s; samplesPerSecond = 2057.6
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.58641615 * 640; EvalErrorPrediction = 0.66250000 * 640; time = 0.1253s; samplesPerSecond = 5106.4
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50844204 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.1526s; samplesPerSecond = 4193.3
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.42228794 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.1795s; samplesPerSecond = 3565.4
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.28812796 * 640; EvalErrorPrediction = 0.59375000 * 640; time = 0.1828s; samplesPerSecond = 3501.8
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.54392000 * 640; EvalErrorPrediction = 0.66718750 * 640; time = 0.1233s; samplesPerSecond = 5188.8
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.29038551 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.3628s; samplesPerSecond = 1764.1
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.27576953 * 640; EvalErrorPrediction = 0.61406250 * 640; time = 0.0981s; samplesPerSecond = 6524.5
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.26696531 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1065s; samplesPerSecond = 6008.9
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.26576222 * 640; EvalErrorPrediction = 0.61093750 * 640; time = 0.1919s; samplesPerSecond = 3335.7
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.06311469 * 20480; EvalErrorPrediction = 0.73945313 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.88529s
|
||||
MPI Rank 1: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62304965 * 640; EvalClassificationError = 0.93437500 * 640; time = 0.2613s; samplesPerSecond = 2449.3
|
||||
MPI Rank 1: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.38132581 * 640; EvalClassificationError = 0.93125000 * 640; time = 0.1965s; samplesPerSecond = 3256.6
|
||||
MPI Rank 1: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98762394 * 640; EvalClassificationError = 0.88593750 * 640; time = 0.4674s; samplesPerSecond = 1369.3
|
||||
MPI Rank 1: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74489685 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1429s; samplesPerSecond = 4478.8
|
||||
MPI Rank 1: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83505115 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.3190s; samplesPerSecond = 2006.1
|
||||
MPI Rank 1: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71847012 * 640; EvalClassificationError = 0.87187500 * 640; time = 0.1405s; samplesPerSecond = 4554.7
|
||||
MPI Rank 1: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.43056226 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.1769s; samplesPerSecond = 3617.6
|
||||
MPI Rank 1: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.55346588 * 640; EvalClassificationError = 0.82656250 * 640; time = 0.3548s; samplesPerSecond = 1803.8
|
||||
MPI Rank 1: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.52102692 * 640; EvalClassificationError = 0.82343750 * 640; time = 0.1120s; samplesPerSecond = 5711.7
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.43553680 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.1256s; samplesPerSecond = 5094.0
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.53125248 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.1788s; samplesPerSecond = 3579.7
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.30802761 * 640; EvalClassificationError = 0.80000000 * 640; time = 0.0993s; samplesPerSecond = 6447.6
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.23431326 * 640; EvalClassificationError = 0.81562500 * 640; time = 0.1337s; samplesPerSecond = 4787.0
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.09809151 * 640; EvalClassificationError = 0.75312500 * 640; time = 0.1556s; samplesPerSecond = 4112.8
|
||||
MPI Rank 1: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.99690035 * 640; EvalClassificationError = 0.74843750 * 640; time = 0.3005s; samplesPerSecond = 2129.6
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.14211613 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.1092s; samplesPerSecond = 5859.1
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.86222858 * 640; EvalClassificationError = 0.72343750 * 640; time = 0.1125s; samplesPerSecond = 5688.0
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.75962647 * 640; EvalClassificationError = 0.65937500 * 640; time = 0.1028s; samplesPerSecond = 6225.1
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.84542338 * 640; EvalClassificationError = 0.72031250 * 640; time = 0.0814s; samplesPerSecond = 7864.6
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.75031878 * 640; EvalClassificationError = 0.68281250 * 640; time = 0.1590s; samplesPerSecond = 4025.9
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.62936500 * 640; EvalClassificationError = 0.66562500 * 640; time = 0.1509s; samplesPerSecond = 4240.3
|
||||
MPI Rank 1: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.62926388 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.1572s; samplesPerSecond = 4071.4
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.55365661 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.3110s; samplesPerSecond = 2057.6
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.58641615 * 640; EvalClassificationError = 0.66250000 * 640; time = 0.1253s; samplesPerSecond = 5106.4
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50844204 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.1526s; samplesPerSecond = 4193.3
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.42228794 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.1795s; samplesPerSecond = 3565.4
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.28812796 * 640; EvalClassificationError = 0.59375000 * 640; time = 0.1828s; samplesPerSecond = 3501.8
|
||||
MPI Rank 1: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.54392000 * 640; EvalClassificationError = 0.66718750 * 640; time = 0.1233s; samplesPerSecond = 5188.8
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.29038551 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.3628s; samplesPerSecond = 1764.1
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.27576953 * 640; EvalClassificationError = 0.61406250 * 640; time = 0.0981s; samplesPerSecond = 6524.5
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.26696531 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1065s; samplesPerSecond = 6008.9
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.26576222 * 640; EvalClassificationError = 0.61093750 * 640; time = 0.1919s; samplesPerSecond = 3335.7
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.06311469 * 20480; EvalClassificationError = 0.73945313 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.88529s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:02: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.16726831 * 2560; EvalErrorPrediction = 0.57851562 * 2560; time = 0.5922s; samplesPerSecond = 4323.0
|
||||
MPI Rank 1: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.08752017 * 2560; EvalErrorPrediction = 0.56523437 * 2560; time = 0.3859s; samplesPerSecond = 6634.3
|
||||
MPI Rank 1: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07134545 * 2560; EvalErrorPrediction = 0.55585938 * 2560; time = 0.6904s; samplesPerSecond = 3707.8
|
||||
MPI Rank 1: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.05824888 * 2560; EvalErrorPrediction = 0.56562500 * 2560; time = 0.3933s; samplesPerSecond = 6508.5
|
||||
MPI Rank 1: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04811411 * 2560; EvalErrorPrediction = 0.55937500 * 2560; time = 0.2376s; samplesPerSecond = 10775.0
|
||||
MPI Rank 1: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.99263458 * 2560; EvalErrorPrediction = 0.54375000 * 2560; time = 0.5335s; samplesPerSecond = 4798.7
|
||||
MPI Rank 1: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99005701 * 2560; EvalErrorPrediction = 0.53359375 * 2560; time = 0.4092s; samplesPerSecond = 6255.5
|
||||
MPI Rank 1: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.97189564 * 2560; EvalErrorPrediction = 0.53984375 * 2560; time = 0.2383s; samplesPerSecond = 10743.7
|
||||
MPI Rank 1: 08/16/2016 10:00:06: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04838552 * 20480; EvalErrorPrediction = 0.55522461 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=3.48694s
|
||||
MPI Rank 1: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.16726831 * 2560; EvalClassificationError = 0.57851562 * 2560; time = 0.5922s; samplesPerSecond = 4323.0
|
||||
MPI Rank 1: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.08752017 * 2560; EvalClassificationError = 0.56523437 * 2560; time = 0.3859s; samplesPerSecond = 6634.3
|
||||
MPI Rank 1: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07134545 * 2560; EvalClassificationError = 0.55585938 * 2560; time = 0.6904s; samplesPerSecond = 3707.8
|
||||
MPI Rank 1: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.05824888 * 2560; EvalClassificationError = 0.56562500 * 2560; time = 0.3933s; samplesPerSecond = 6508.5
|
||||
MPI Rank 1: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04811411 * 2560; EvalClassificationError = 0.55937500 * 2560; time = 0.2376s; samplesPerSecond = 10775.0
|
||||
MPI Rank 1: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.99263458 * 2560; EvalClassificationError = 0.54375000 * 2560; time = 0.5335s; samplesPerSecond = 4798.7
|
||||
MPI Rank 1: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99005701 * 2560; EvalClassificationError = 0.53359375 * 2560; time = 0.4092s; samplesPerSecond = 6255.5
|
||||
MPI Rank 1: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.97189564 * 2560; EvalClassificationError = 0.53984375 * 2560; time = 0.2383s; samplesPerSecond = 10743.7
|
||||
MPI Rank 1: 08/16/2016 10:00:06: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04838552 * 20480; EvalClassificationError = 0.55522461 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=3.48694s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:06: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:06: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:00:07: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97114275 * 10240; EvalErrorPrediction = 0.53769531 * 10240; time = 0.9880s; samplesPerSecond = 10364.1
|
||||
MPI Rank 1: 08/16/2016 10:00:08: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.99390313 * 10240; EvalErrorPrediction = 0.55712891 * 10240; time = 1.4042s; samplesPerSecond = 7292.5
|
||||
MPI Rank 1: 08/16/2016 10:00:08: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.98252294 * 20480; EvalErrorPrediction = 0.54741211 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.39895s
|
||||
MPI Rank 1: 08/16/2016 10:00:07: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97114275 * 10240; EvalClassificationError = 0.53769531 * 10240; time = 0.9880s; samplesPerSecond = 10364.1
|
||||
MPI Rank 1: 08/16/2016 10:00:08: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.99390313 * 10240; EvalClassificationError = 0.55712891 * 10240; time = 1.4042s; samplesPerSecond = 7292.5
|
||||
MPI Rank 1: 08/16/2016 10:00:08: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.98252294 * 20480; EvalClassificationError = 0.54741211 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.39895s
|
||||
MPI Rank 1: 08/16/2016 10:00:08: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:08: Action "train" complete.
|
||||
|
|
@ -1226,7 +1226,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1252,7 +1252,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1329,7 +1329,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1355,7 +1355,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1438,7 +1438,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1464,7 +1464,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1560,7 +1560,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1589,7 +1589,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1612,7 +1612,7 @@ MPI Rank 2: 08/16/2016 09:59:55: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 09:59:55: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:59:55: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 09:59:55: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 09:59:55: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1665,61 +1665,61 @@ MPI Rank 2: 08/16/2016 09:59:56: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:59:56: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62304965 * 640; EvalErrorPrediction = 0.93437500 * 640; time = 0.2618s; samplesPerSecond = 2444.4
|
||||
MPI Rank 2: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.38132581 * 640; EvalErrorPrediction = 0.93125000 * 640; time = 0.1965s; samplesPerSecond = 3256.9
|
||||
MPI Rank 2: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98762394 * 640; EvalErrorPrediction = 0.88593750 * 640; time = 0.4698s; samplesPerSecond = 1362.4
|
||||
MPI Rank 2: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74489685 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1451s; samplesPerSecond = 4409.9
|
||||
MPI Rank 2: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83505115 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.3168s; samplesPerSecond = 2020.3
|
||||
MPI Rank 2: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71847012 * 640; EvalErrorPrediction = 0.87187500 * 640; time = 0.1378s; samplesPerSecond = 4643.7
|
||||
MPI Rank 2: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.43056226 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.1768s; samplesPerSecond = 3619.3
|
||||
MPI Rank 2: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.55346588 * 640; EvalErrorPrediction = 0.82656250 * 640; time = 0.3551s; samplesPerSecond = 1802.5
|
||||
MPI Rank 2: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.52102692 * 640; EvalErrorPrediction = 0.82343750 * 640; time = 0.1120s; samplesPerSecond = 5716.4
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.43553680 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.1254s; samplesPerSecond = 5103.1
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.53125248 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.1793s; samplesPerSecond = 3570.1
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.30802761 * 640; EvalErrorPrediction = 0.80000000 * 640; time = 0.1012s; samplesPerSecond = 6324.6
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.23431326 * 640; EvalErrorPrediction = 0.81562500 * 640; time = 0.1353s; samplesPerSecond = 4730.5
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.09809151 * 640; EvalErrorPrediction = 0.75312500 * 640; time = 0.1533s; samplesPerSecond = 4174.6
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.99690035 * 640; EvalErrorPrediction = 0.74843750 * 640; time = 0.2988s; samplesPerSecond = 2142.0
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.14211613 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.1116s; samplesPerSecond = 5734.6
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.86222858 * 640; EvalErrorPrediction = 0.72343750 * 640; time = 0.1125s; samplesPerSecond = 5688.6
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.75962647 * 640; EvalErrorPrediction = 0.65937500 * 640; time = 0.1004s; samplesPerSecond = 6377.4
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.84542338 * 640; EvalErrorPrediction = 0.72031250 * 640; time = 0.0814s; samplesPerSecond = 7865.6
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.75031878 * 640; EvalErrorPrediction = 0.68281250 * 640; time = 0.1589s; samplesPerSecond = 4028.4
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.62936500 * 640; EvalErrorPrediction = 0.66562500 * 640; time = 0.1510s; samplesPerSecond = 4238.3
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.62926388 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.1573s; samplesPerSecond = 4068.3
|
||||
MPI Rank 2: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.55365661 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.3108s; samplesPerSecond = 2059.0
|
||||
MPI Rank 2: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.58641615 * 640; EvalErrorPrediction = 0.66250000 * 640; time = 0.1262s; samplesPerSecond = 5072.7
|
||||
MPI Rank 2: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50844204 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.1518s; samplesPerSecond = 4216.4
|
||||
MPI Rank 2: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.42228794 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.1793s; samplesPerSecond = 3568.9
|
||||
MPI Rank 2: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.28812796 * 640; EvalErrorPrediction = 0.59375000 * 640; time = 0.1830s; samplesPerSecond = 3497.1
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.54392000 * 640; EvalErrorPrediction = 0.66718750 * 640; time = 0.2486s; samplesPerSecond = 2574.6
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.29038551 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.2396s; samplesPerSecond = 2671.5
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.27576953 * 640; EvalErrorPrediction = 0.61406250 * 640; time = 0.0959s; samplesPerSecond = 6676.2
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.26696531 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1070s; samplesPerSecond = 5980.8
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.26576222 * 640; EvalErrorPrediction = 0.61093750 * 640; time = 0.1915s; samplesPerSecond = 3341.8
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.06311469 * 20480; EvalErrorPrediction = 0.73945313 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.88536s
|
||||
MPI Rank 2: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62304965 * 640; EvalClassificationError = 0.93437500 * 640; time = 0.2618s; samplesPerSecond = 2444.4
|
||||
MPI Rank 2: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.38132581 * 640; EvalClassificationError = 0.93125000 * 640; time = 0.1965s; samplesPerSecond = 3256.9
|
||||
MPI Rank 2: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98762394 * 640; EvalClassificationError = 0.88593750 * 640; time = 0.4698s; samplesPerSecond = 1362.4
|
||||
MPI Rank 2: 08/16/2016 09:59:57: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74489685 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1451s; samplesPerSecond = 4409.9
|
||||
MPI Rank 2: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83505115 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.3168s; samplesPerSecond = 2020.3
|
||||
MPI Rank 2: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71847012 * 640; EvalClassificationError = 0.87187500 * 640; time = 0.1378s; samplesPerSecond = 4643.7
|
||||
MPI Rank 2: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.43056226 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.1768s; samplesPerSecond = 3619.3
|
||||
MPI Rank 2: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.55346588 * 640; EvalClassificationError = 0.82656250 * 640; time = 0.3551s; samplesPerSecond = 1802.5
|
||||
MPI Rank 2: 08/16/2016 09:59:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.52102692 * 640; EvalClassificationError = 0.82343750 * 640; time = 0.1120s; samplesPerSecond = 5716.4
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.43553680 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.1254s; samplesPerSecond = 5103.1
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.53125248 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.1793s; samplesPerSecond = 3570.1
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.30802761 * 640; EvalClassificationError = 0.80000000 * 640; time = 0.1012s; samplesPerSecond = 6324.6
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.23431326 * 640; EvalClassificationError = 0.81562500 * 640; time = 0.1353s; samplesPerSecond = 4730.5
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.09809151 * 640; EvalClassificationError = 0.75312500 * 640; time = 0.1533s; samplesPerSecond = 4174.6
|
||||
MPI Rank 2: 08/16/2016 09:59:59: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.99690035 * 640; EvalClassificationError = 0.74843750 * 640; time = 0.2988s; samplesPerSecond = 2142.0
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.14211613 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.1116s; samplesPerSecond = 5734.6
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.86222858 * 640; EvalClassificationError = 0.72343750 * 640; time = 0.1125s; samplesPerSecond = 5688.6
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.75962647 * 640; EvalClassificationError = 0.65937500 * 640; time = 0.1004s; samplesPerSecond = 6377.4
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.84542338 * 640; EvalClassificationError = 0.72031250 * 640; time = 0.0814s; samplesPerSecond = 7865.6
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.75031878 * 640; EvalClassificationError = 0.68281250 * 640; time = 0.1589s; samplesPerSecond = 4028.4
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.62936500 * 640; EvalClassificationError = 0.66562500 * 640; time = 0.1510s; samplesPerSecond = 4238.3
|
||||
MPI Rank 2: 08/16/2016 10:00:00: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.62926388 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.1573s; samplesPerSecond = 4068.3
|
||||
MPI Rank 2: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.55365661 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.3108s; samplesPerSecond = 2059.0
|
||||
MPI Rank 2: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.58641615 * 640; EvalClassificationError = 0.66250000 * 640; time = 0.1262s; samplesPerSecond = 5072.7
|
||||
MPI Rank 2: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50844204 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.1518s; samplesPerSecond = 4216.4
|
||||
MPI Rank 2: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.42228794 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.1793s; samplesPerSecond = 3568.9
|
||||
MPI Rank 2: 08/16/2016 10:00:01: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.28812796 * 640; EvalClassificationError = 0.59375000 * 640; time = 0.1830s; samplesPerSecond = 3497.1
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.54392000 * 640; EvalClassificationError = 0.66718750 * 640; time = 0.2486s; samplesPerSecond = 2574.6
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.29038551 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.2396s; samplesPerSecond = 2671.5
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.27576953 * 640; EvalClassificationError = 0.61406250 * 640; time = 0.0959s; samplesPerSecond = 6676.2
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.26696531 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1070s; samplesPerSecond = 5980.8
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.26576222 * 640; EvalClassificationError = 0.61093750 * 640; time = 0.1915s; samplesPerSecond = 3341.8
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.06311469 * 20480; EvalClassificationError = 0.73945313 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=5.88536s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:02: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.16726831 * 2560; EvalErrorPrediction = 0.57851562 * 2560; time = 0.5955s; samplesPerSecond = 4298.6
|
||||
MPI Rank 2: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.08752017 * 2560; EvalErrorPrediction = 0.56523437 * 2560; time = 0.3830s; samplesPerSecond = 6684.6
|
||||
MPI Rank 2: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07134545 * 2560; EvalErrorPrediction = 0.55585938 * 2560; time = 0.6922s; samplesPerSecond = 3698.2
|
||||
MPI Rank 2: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.05824888 * 2560; EvalErrorPrediction = 0.56562500 * 2560; time = 0.3959s; samplesPerSecond = 6466.3
|
||||
MPI Rank 2: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04811411 * 2560; EvalErrorPrediction = 0.55937500 * 2560; time = 0.2337s; samplesPerSecond = 10954.3
|
||||
MPI Rank 2: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.99263458 * 2560; EvalErrorPrediction = 0.54375000 * 2560; time = 0.5339s; samplesPerSecond = 4794.6
|
||||
MPI Rank 2: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99005701 * 2560; EvalErrorPrediction = 0.53359375 * 2560; time = 0.4092s; samplesPerSecond = 6255.7
|
||||
MPI Rank 2: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.97189564 * 2560; EvalErrorPrediction = 0.53984375 * 2560; time = 0.2400s; samplesPerSecond = 10667.6
|
||||
MPI Rank 2: 08/16/2016 10:00:06: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04838552 * 20480; EvalErrorPrediction = 0.55522461 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=3.48698s
|
||||
MPI Rank 2: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.16726831 * 2560; EvalClassificationError = 0.57851562 * 2560; time = 0.5955s; samplesPerSecond = 4298.6
|
||||
MPI Rank 2: 08/16/2016 10:00:03: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.08752017 * 2560; EvalClassificationError = 0.56523437 * 2560; time = 0.3830s; samplesPerSecond = 6684.6
|
||||
MPI Rank 2: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.07134545 * 2560; EvalClassificationError = 0.55585938 * 2560; time = 0.6922s; samplesPerSecond = 3698.2
|
||||
MPI Rank 2: 08/16/2016 10:00:04: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.05824888 * 2560; EvalClassificationError = 0.56562500 * 2560; time = 0.3959s; samplesPerSecond = 6466.3
|
||||
MPI Rank 2: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.04811411 * 2560; EvalClassificationError = 0.55937500 * 2560; time = 0.2337s; samplesPerSecond = 10954.3
|
||||
MPI Rank 2: 08/16/2016 10:00:05: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.99263458 * 2560; EvalClassificationError = 0.54375000 * 2560; time = 0.5339s; samplesPerSecond = 4794.6
|
||||
MPI Rank 2: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.99005701 * 2560; EvalClassificationError = 0.53359375 * 2560; time = 0.4092s; samplesPerSecond = 6255.7
|
||||
MPI Rank 2: 08/16/2016 10:00:06: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.97189564 * 2560; EvalClassificationError = 0.53984375 * 2560; time = 0.2400s; samplesPerSecond = 10667.6
|
||||
MPI Rank 2: 08/16/2016 10:00:06: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 2.04838552 * 20480; EvalClassificationError = 0.55522461 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=3.48698s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:06: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:06: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 10:00:07: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97114275 * 10240; EvalErrorPrediction = 0.53769531 * 10240; time = 0.9897s; samplesPerSecond = 10346.1
|
||||
MPI Rank 2: 08/16/2016 10:00:08: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.99390313 * 10240; EvalErrorPrediction = 0.55712891 * 10240; time = 1.4041s; samplesPerSecond = 7292.7
|
||||
MPI Rank 2: 08/16/2016 10:00:08: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.98252294 * 20480; EvalErrorPrediction = 0.54741211 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.39757s
|
||||
MPI Rank 2: 08/16/2016 10:00:07: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.97114275 * 10240; EvalClassificationError = 0.53769531 * 10240; time = 0.9897s; samplesPerSecond = 10346.1
|
||||
MPI Rank 2: 08/16/2016 10:00:08: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.99390313 * 10240; EvalClassificationError = 0.55712891 * 10240; time = 1.4041s; samplesPerSecond = 7292.7
|
||||
MPI Rank 2: 08/16/2016 10:00:08: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.98252294 * 20480; EvalClassificationError = 0.54741211 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=2.39757s
|
||||
MPI Rank 2: 08/16/2016 10:00:08: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:08: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -133,7 +133,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -159,7 +159,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -236,7 +236,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -262,7 +262,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -345,7 +345,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -371,7 +371,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -468,7 +468,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -497,7 +497,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -520,7 +520,7 @@ MPI Rank 0: 08/16/2016 10:00:13: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 10:00:13: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:00:13: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 10:00:13: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 10:00:13: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -573,63 +573,63 @@ MPI Rank 0: 08/16/2016 10:00:15: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalErrorPrediction = 0.92031250 * 640; time = 0.0693s; samplesPerSecond = 9238.4
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.0567s; samplesPerSecond = 11281.5
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalErrorPrediction = 0.87187500 * 640; time = 0.0566s; samplesPerSecond = 11312.8
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0594s; samplesPerSecond = 10775.7
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0581s; samplesPerSecond = 11023.1
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.0572s; samplesPerSecond = 11196.6
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0570s; samplesPerSecond = 11224.7
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalErrorPrediction = 0.82968750 * 640; time = 0.0567s; samplesPerSecond = 11282.7
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalErrorPrediction = 0.81093750 * 640; time = 0.0569s; samplesPerSecond = 11247.6
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.0567s; samplesPerSecond = 11294.4
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0570s; samplesPerSecond = 11235.8
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalErrorPrediction = 0.75625000 * 640; time = 0.0568s; samplesPerSecond = 11264.2
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalErrorPrediction = 0.75312500 * 640; time = 0.0569s; samplesPerSecond = 11251.2
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0574s; samplesPerSecond = 11155.1
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalErrorPrediction = 0.70000000 * 640; time = 0.0573s; samplesPerSecond = 11177.5
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalErrorPrediction = 0.74218750 * 640; time = 0.0572s; samplesPerSecond = 11195.9
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalErrorPrediction = 0.69375000 * 640; time = 0.0570s; samplesPerSecond = 11226.1
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0568s; samplesPerSecond = 11271.6
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalErrorPrediction = 0.69062500 * 640; time = 0.0571s; samplesPerSecond = 11201.1
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalErrorPrediction = 0.68593750 * 640; time = 0.0567s; samplesPerSecond = 11290.1
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalErrorPrediction = 0.64375000 * 640; time = 0.0570s; samplesPerSecond = 11237.9
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0577s; samplesPerSecond = 11100.3
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0570s; samplesPerSecond = 11218.2
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.0572s; samplesPerSecond = 11183.5
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalErrorPrediction = 0.63125000 * 640; time = 0.0570s; samplesPerSecond = 11221.4
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalErrorPrediction = 0.61562500 * 640; time = 0.0566s; samplesPerSecond = 11311.6
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0566s; samplesPerSecond = 11298.8
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0569s; samplesPerSecond = 11254.9
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0567s; samplesPerSecond = 11287.3
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalErrorPrediction = 0.60781250 * 640; time = 0.0568s; samplesPerSecond = 11275.0
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0569s; samplesPerSecond = 11244.2
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalErrorPrediction = 0.58593750 * 640; time = 0.0566s; samplesPerSecond = 11303.6
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalErrorPrediction = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.84563s
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalClassificationError = 0.92031250 * 640; time = 0.0693s; samplesPerSecond = 9238.4
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.0567s; samplesPerSecond = 11281.5
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalClassificationError = 0.87187500 * 640; time = 0.0566s; samplesPerSecond = 11312.8
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0594s; samplesPerSecond = 10775.7
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0581s; samplesPerSecond = 11023.1
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.0572s; samplesPerSecond = 11196.6
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0570s; samplesPerSecond = 11224.7
|
||||
MPI Rank 0: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalClassificationError = 0.82968750 * 640; time = 0.0567s; samplesPerSecond = 11282.7
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalClassificationError = 0.81093750 * 640; time = 0.0569s; samplesPerSecond = 11247.6
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.0567s; samplesPerSecond = 11294.4
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0570s; samplesPerSecond = 11235.8
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalClassificationError = 0.75625000 * 640; time = 0.0568s; samplesPerSecond = 11264.2
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalClassificationError = 0.75312500 * 640; time = 0.0569s; samplesPerSecond = 11251.2
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0574s; samplesPerSecond = 11155.1
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalClassificationError = 0.70000000 * 640; time = 0.0573s; samplesPerSecond = 11177.5
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalClassificationError = 0.74218750 * 640; time = 0.0572s; samplesPerSecond = 11195.9
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalClassificationError = 0.69375000 * 640; time = 0.0570s; samplesPerSecond = 11226.1
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0568s; samplesPerSecond = 11271.6
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalClassificationError = 0.69062500 * 640; time = 0.0571s; samplesPerSecond = 11201.1
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalClassificationError = 0.68593750 * 640; time = 0.0567s; samplesPerSecond = 11290.1
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalClassificationError = 0.64375000 * 640; time = 0.0570s; samplesPerSecond = 11237.9
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0577s; samplesPerSecond = 11100.3
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0570s; samplesPerSecond = 11218.2
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.0572s; samplesPerSecond = 11183.5
|
||||
MPI Rank 0: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalClassificationError = 0.63125000 * 640; time = 0.0570s; samplesPerSecond = 11221.4
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalClassificationError = 0.61562500 * 640; time = 0.0566s; samplesPerSecond = 11311.6
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0566s; samplesPerSecond = 11298.8
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0569s; samplesPerSecond = 11254.9
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0567s; samplesPerSecond = 11287.3
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalClassificationError = 0.60781250 * 640; time = 0.0568s; samplesPerSecond = 11275.0
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0569s; samplesPerSecond = 11244.2
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalClassificationError = 0.58593750 * 640; time = 0.0566s; samplesPerSecond = 11303.6
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalClassificationError = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.84563s
|
||||
MPI Rank 0: 08/16/2016 10:00:17: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelNoQuantization@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalErrorPrediction = 0.56367188 * 2560; time = 0.0612s; samplesPerSecond = 41819.8
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalErrorPrediction = 0.54218750 * 2560; time = 0.0599s; samplesPerSecond = 42761.5
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalErrorPrediction = 0.54257813 * 2560; time = 0.0599s; samplesPerSecond = 42762.2
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalErrorPrediction = 0.54648438 * 2560; time = 0.0596s; samplesPerSecond = 42934.3
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalErrorPrediction = 0.53984375 * 2560; time = 0.0606s; samplesPerSecond = 42238.0
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalErrorPrediction = 0.52109375 * 2560; time = 0.0582s; samplesPerSecond = 43974.2
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalErrorPrediction = 0.52148438 * 2560; time = 0.0596s; samplesPerSecond = 42973.9
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalErrorPrediction = 0.51992187 * 2560; time = 0.0593s; samplesPerSecond = 43139.0
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalErrorPrediction = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.483662s
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalClassificationError = 0.56367188 * 2560; time = 0.0612s; samplesPerSecond = 41819.8
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalClassificationError = 0.54218750 * 2560; time = 0.0599s; samplesPerSecond = 42761.5
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalClassificationError = 0.54257813 * 2560; time = 0.0599s; samplesPerSecond = 42762.2
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalClassificationError = 0.54648438 * 2560; time = 0.0596s; samplesPerSecond = 42934.3
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalClassificationError = 0.53984375 * 2560; time = 0.0606s; samplesPerSecond = 42238.0
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalClassificationError = 0.52109375 * 2560; time = 0.0582s; samplesPerSecond = 43974.2
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalClassificationError = 0.52148438 * 2560; time = 0.0596s; samplesPerSecond = 42973.9
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalClassificationError = 0.51992187 * 2560; time = 0.0593s; samplesPerSecond = 43139.0
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalClassificationError = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.483662s
|
||||
MPI Rank 0: 08/16/2016 10:00:17: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelNoQuantization@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 10:00:17: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalErrorPrediction = 0.52470703 * 10240; time = 0.0894s; samplesPerSecond = 114526.0
|
||||
MPI Rank 0: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalErrorPrediction = 0.53974609 * 10240; time = 0.0835s; samplesPerSecond = 122653.8
|
||||
MPI Rank 0: 08/16/2016 10:00:18: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalErrorPrediction = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.179159s
|
||||
MPI Rank 0: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalClassificationError = 0.52470703 * 10240; time = 0.0894s; samplesPerSecond = 114526.0
|
||||
MPI Rank 0: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalClassificationError = 0.53974609 * 10240; time = 0.0835s; samplesPerSecond = 122653.8
|
||||
MPI Rank 0: 08/16/2016 10:00:18: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalClassificationError = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.179159s
|
||||
MPI Rank 0: 08/16/2016 10:00:18: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095705.492453/Speech/DNN_ParallelNoQuantization@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 10:00:18: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -682,7 +682,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -708,7 +708,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -785,7 +785,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -811,7 +811,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -894,7 +894,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -920,7 +920,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1017,7 +1017,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1046,7 +1046,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1069,7 +1069,7 @@ MPI Rank 1: 08/16/2016 10:00:13: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 10:00:13: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:13: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 10:00:13: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 10:00:13: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1122,61 +1122,61 @@ MPI Rank 1: 08/16/2016 10:00:15: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalErrorPrediction = 0.92031250 * 640; time = 0.0695s; samplesPerSecond = 9204.8
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.0563s; samplesPerSecond = 11359.2
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalErrorPrediction = 0.87187500 * 640; time = 0.0567s; samplesPerSecond = 11286.5
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0593s; samplesPerSecond = 10788.9
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0582s; samplesPerSecond = 10994.1
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.0572s; samplesPerSecond = 11191.6
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0571s; samplesPerSecond = 11217.8
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalErrorPrediction = 0.82968750 * 640; time = 0.0567s; samplesPerSecond = 11279.9
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalErrorPrediction = 0.81093750 * 640; time = 0.0568s; samplesPerSecond = 11262.3
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.0571s; samplesPerSecond = 11212.5
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0567s; samplesPerSecond = 11283.7
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalErrorPrediction = 0.75625000 * 640; time = 0.0571s; samplesPerSecond = 11201.9
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalErrorPrediction = 0.75312500 * 640; time = 0.0569s; samplesPerSecond = 11241.9
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0570s; samplesPerSecond = 11230.0
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalErrorPrediction = 0.70000000 * 640; time = 0.0573s; samplesPerSecond = 11176.5
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalErrorPrediction = 0.74218750 * 640; time = 0.0572s; samplesPerSecond = 11194.3
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalErrorPrediction = 0.69375000 * 640; time = 0.0570s; samplesPerSecond = 11224.7
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0572s; samplesPerSecond = 11191.0
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalErrorPrediction = 0.69062500 * 640; time = 0.0572s; samplesPerSecond = 11195.9
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalErrorPrediction = 0.68593750 * 640; time = 0.0567s; samplesPerSecond = 11279.7
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalErrorPrediction = 0.64375000 * 640; time = 0.0570s; samplesPerSecond = 11233.0
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0573s; samplesPerSecond = 11172.4
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0571s; samplesPerSecond = 11216.7
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.0573s; samplesPerSecond = 11161.9
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalErrorPrediction = 0.63125000 * 640; time = 0.0571s; samplesPerSecond = 11209.8
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalErrorPrediction = 0.61562500 * 640; time = 0.0566s; samplesPerSecond = 11303.4
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0570s; samplesPerSecond = 11236.9
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0564s; samplesPerSecond = 11344.5
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0568s; samplesPerSecond = 11271.8
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalErrorPrediction = 0.60781250 * 640; time = 0.0569s; samplesPerSecond = 11248.2
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0570s; samplesPerSecond = 11234.2
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalErrorPrediction = 0.58593750 * 640; time = 0.0567s; samplesPerSecond = 11296.2
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalErrorPrediction = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.84538s
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalClassificationError = 0.92031250 * 640; time = 0.0695s; samplesPerSecond = 9204.8
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.0563s; samplesPerSecond = 11359.2
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalClassificationError = 0.87187500 * 640; time = 0.0567s; samplesPerSecond = 11286.5
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0593s; samplesPerSecond = 10788.9
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0582s; samplesPerSecond = 10994.1
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.0572s; samplesPerSecond = 11191.6
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0571s; samplesPerSecond = 11217.8
|
||||
MPI Rank 1: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalClassificationError = 0.82968750 * 640; time = 0.0567s; samplesPerSecond = 11279.9
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalClassificationError = 0.81093750 * 640; time = 0.0568s; samplesPerSecond = 11262.3
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.0571s; samplesPerSecond = 11212.5
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0567s; samplesPerSecond = 11283.7
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalClassificationError = 0.75625000 * 640; time = 0.0571s; samplesPerSecond = 11201.9
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalClassificationError = 0.75312500 * 640; time = 0.0569s; samplesPerSecond = 11241.9
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0570s; samplesPerSecond = 11230.0
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalClassificationError = 0.70000000 * 640; time = 0.0573s; samplesPerSecond = 11176.5
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalClassificationError = 0.74218750 * 640; time = 0.0572s; samplesPerSecond = 11194.3
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalClassificationError = 0.69375000 * 640; time = 0.0570s; samplesPerSecond = 11224.7
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0572s; samplesPerSecond = 11191.0
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalClassificationError = 0.69062500 * 640; time = 0.0572s; samplesPerSecond = 11195.9
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalClassificationError = 0.68593750 * 640; time = 0.0567s; samplesPerSecond = 11279.7
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalClassificationError = 0.64375000 * 640; time = 0.0570s; samplesPerSecond = 11233.0
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0573s; samplesPerSecond = 11172.4
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0571s; samplesPerSecond = 11216.7
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.0573s; samplesPerSecond = 11161.9
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalClassificationError = 0.63125000 * 640; time = 0.0571s; samplesPerSecond = 11209.8
|
||||
MPI Rank 1: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalClassificationError = 0.61562500 * 640; time = 0.0566s; samplesPerSecond = 11303.4
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0570s; samplesPerSecond = 11236.9
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0564s; samplesPerSecond = 11344.5
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0568s; samplesPerSecond = 11271.8
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalClassificationError = 0.60781250 * 640; time = 0.0569s; samplesPerSecond = 11248.2
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0570s; samplesPerSecond = 11234.2
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalClassificationError = 0.58593750 * 640; time = 0.0567s; samplesPerSecond = 11296.2
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalClassificationError = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.84538s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalErrorPrediction = 0.56367188 * 2560; time = 0.0614s; samplesPerSecond = 41699.9
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalErrorPrediction = 0.54218750 * 2560; time = 0.0597s; samplesPerSecond = 42846.6
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalErrorPrediction = 0.54257813 * 2560; time = 0.0599s; samplesPerSecond = 42717.2
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalErrorPrediction = 0.54648438 * 2560; time = 0.0600s; samplesPerSecond = 42649.6
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalErrorPrediction = 0.53984375 * 2560; time = 0.0602s; samplesPerSecond = 42496.7
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalErrorPrediction = 0.52109375 * 2560; time = 0.0583s; samplesPerSecond = 43874.7
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalErrorPrediction = 0.52148438 * 2560; time = 0.0595s; samplesPerSecond = 43039.7
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalErrorPrediction = 0.51992187 * 2560; time = 0.0595s; samplesPerSecond = 43045.5
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalErrorPrediction = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.483806s
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalClassificationError = 0.56367188 * 2560; time = 0.0614s; samplesPerSecond = 41699.9
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalClassificationError = 0.54218750 * 2560; time = 0.0597s; samplesPerSecond = 42846.6
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalClassificationError = 0.54257813 * 2560; time = 0.0599s; samplesPerSecond = 42717.2
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalClassificationError = 0.54648438 * 2560; time = 0.0600s; samplesPerSecond = 42649.6
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalClassificationError = 0.53984375 * 2560; time = 0.0602s; samplesPerSecond = 42496.7
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalClassificationError = 0.52109375 * 2560; time = 0.0583s; samplesPerSecond = 43874.7
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalClassificationError = 0.52148438 * 2560; time = 0.0595s; samplesPerSecond = 43039.7
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalClassificationError = 0.51992187 * 2560; time = 0.0595s; samplesPerSecond = 43045.5
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalClassificationError = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.483806s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:17: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalErrorPrediction = 0.52470703 * 10240; time = 0.0900s; samplesPerSecond = 113825.8
|
||||
MPI Rank 1: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalErrorPrediction = 0.53974609 * 10240; time = 0.0832s; samplesPerSecond = 123133.2
|
||||
MPI Rank 1: 08/16/2016 10:00:18: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalErrorPrediction = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.179295s
|
||||
MPI Rank 1: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalClassificationError = 0.52470703 * 10240; time = 0.0900s; samplesPerSecond = 113825.8
|
||||
MPI Rank 1: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalClassificationError = 0.53974609 * 10240; time = 0.0832s; samplesPerSecond = 123133.2
|
||||
MPI Rank 1: 08/16/2016 10:00:18: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalClassificationError = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.179295s
|
||||
MPI Rank 1: 08/16/2016 10:00:18: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 10:00:18: Action "train" complete.
|
||||
|
|
@ -1228,7 +1228,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1254,7 +1254,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1331,7 +1331,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1357,7 +1357,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1440,7 +1440,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1466,7 +1466,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1563,7 +1563,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1592,7 +1592,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1615,7 +1615,7 @@ MPI Rank 2: 08/16/2016 10:00:14: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 10:00:14: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:14: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 10:00:14: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 10:00:14: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1668,61 +1668,61 @@ MPI Rank 2: 08/16/2016 10:00:15: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalErrorPrediction = 0.92031250 * 640; time = 0.0697s; samplesPerSecond = 9180.2
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.0565s; samplesPerSecond = 11323.4
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalErrorPrediction = 0.87187500 * 640; time = 0.0568s; samplesPerSecond = 11270.0
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0594s; samplesPerSecond = 10778.6
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0581s; samplesPerSecond = 11010.4
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.0572s; samplesPerSecond = 11196.2
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0570s; samplesPerSecond = 11220.8
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalErrorPrediction = 0.82968750 * 640; time = 0.0567s; samplesPerSecond = 11278.7
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalErrorPrediction = 0.81093750 * 640; time = 0.0567s; samplesPerSecond = 11290.1
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.0569s; samplesPerSecond = 11251.6
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0570s; samplesPerSecond = 11230.8
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalErrorPrediction = 0.75625000 * 640; time = 0.0568s; samplesPerSecond = 11259.5
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalErrorPrediction = 0.75312500 * 640; time = 0.0569s; samplesPerSecond = 11245.0
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0572s; samplesPerSecond = 11198.0
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalErrorPrediction = 0.70000000 * 640; time = 0.0572s; samplesPerSecond = 11183.3
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalErrorPrediction = 0.74218750 * 640; time = 0.0571s; samplesPerSecond = 11201.0
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalErrorPrediction = 0.69375000 * 640; time = 0.0570s; samplesPerSecond = 11231.8
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0570s; samplesPerSecond = 11229.3
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalErrorPrediction = 0.69062500 * 640; time = 0.0572s; samplesPerSecond = 11196.2
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalErrorPrediction = 0.68593750 * 640; time = 0.0567s; samplesPerSecond = 11284.3
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalErrorPrediction = 0.64375000 * 640; time = 0.0570s; samplesPerSecond = 11233.8
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0574s; samplesPerSecond = 11145.7
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0570s; samplesPerSecond = 11224.1
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.0574s; samplesPerSecond = 11142.8
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalErrorPrediction = 0.63125000 * 640; time = 0.0571s; samplesPerSecond = 11215.1
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalErrorPrediction = 0.61562500 * 640; time = 0.0566s; samplesPerSecond = 11306.4
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0566s; samplesPerSecond = 11298.0
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0566s; samplesPerSecond = 11310.6
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0567s; samplesPerSecond = 11282.5
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalErrorPrediction = 0.60781250 * 640; time = 0.0570s; samplesPerSecond = 11227.9
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0569s; samplesPerSecond = 11240.9
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalErrorPrediction = 0.58593750 * 640; time = 0.0566s; samplesPerSecond = 11299.0
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalErrorPrediction = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.84551s
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalClassificationError = 0.92031250 * 640; time = 0.0697s; samplesPerSecond = 9180.2
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.0565s; samplesPerSecond = 11323.4
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalClassificationError = 0.87187500 * 640; time = 0.0568s; samplesPerSecond = 11270.0
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0594s; samplesPerSecond = 10778.6
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0581s; samplesPerSecond = 11010.4
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.0572s; samplesPerSecond = 11196.2
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0570s; samplesPerSecond = 11220.8
|
||||
MPI Rank 2: 08/16/2016 10:00:15: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalClassificationError = 0.82968750 * 640; time = 0.0567s; samplesPerSecond = 11278.7
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalClassificationError = 0.81093750 * 640; time = 0.0567s; samplesPerSecond = 11290.1
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.0569s; samplesPerSecond = 11251.6
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0570s; samplesPerSecond = 11230.8
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalClassificationError = 0.75625000 * 640; time = 0.0568s; samplesPerSecond = 11259.5
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalClassificationError = 0.75312500 * 640; time = 0.0569s; samplesPerSecond = 11245.0
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0572s; samplesPerSecond = 11198.0
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalClassificationError = 0.70000000 * 640; time = 0.0572s; samplesPerSecond = 11183.3
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalClassificationError = 0.74218750 * 640; time = 0.0571s; samplesPerSecond = 11201.0
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalClassificationError = 0.69375000 * 640; time = 0.0570s; samplesPerSecond = 11231.8
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0570s; samplesPerSecond = 11229.3
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalClassificationError = 0.69062500 * 640; time = 0.0572s; samplesPerSecond = 11196.2
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalClassificationError = 0.68593750 * 640; time = 0.0567s; samplesPerSecond = 11284.3
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalClassificationError = 0.64375000 * 640; time = 0.0570s; samplesPerSecond = 11233.8
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0574s; samplesPerSecond = 11145.7
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0570s; samplesPerSecond = 11224.1
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.0574s; samplesPerSecond = 11142.8
|
||||
MPI Rank 2: 08/16/2016 10:00:16: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalClassificationError = 0.63125000 * 640; time = 0.0571s; samplesPerSecond = 11215.1
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalClassificationError = 0.61562500 * 640; time = 0.0566s; samplesPerSecond = 11306.4
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0566s; samplesPerSecond = 11298.0
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0566s; samplesPerSecond = 11310.6
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0567s; samplesPerSecond = 11282.5
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalClassificationError = 0.60781250 * 640; time = 0.0570s; samplesPerSecond = 11227.9
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0569s; samplesPerSecond = 11240.9
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalClassificationError = 0.58593750 * 640; time = 0.0566s; samplesPerSecond = 11299.0
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalClassificationError = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.84551s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalErrorPrediction = 0.56367188 * 2560; time = 0.0612s; samplesPerSecond = 41826.0
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalErrorPrediction = 0.54218750 * 2560; time = 0.0597s; samplesPerSecond = 42906.2
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalErrorPrediction = 0.54257813 * 2560; time = 0.0601s; samplesPerSecond = 42627.6
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalErrorPrediction = 0.54648438 * 2560; time = 0.0597s; samplesPerSecond = 42909.1
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalErrorPrediction = 0.53984375 * 2560; time = 0.0604s; samplesPerSecond = 42394.6
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalErrorPrediction = 0.52109375 * 2560; time = 0.0584s; samplesPerSecond = 43804.9
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalErrorPrediction = 0.52148438 * 2560; time = 0.0593s; samplesPerSecond = 43149.9
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalErrorPrediction = 0.51992187 * 2560; time = 0.0596s; samplesPerSecond = 42978.3
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalErrorPrediction = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.483777s
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalClassificationError = 0.56367188 * 2560; time = 0.0612s; samplesPerSecond = 41826.0
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalClassificationError = 0.54218750 * 2560; time = 0.0597s; samplesPerSecond = 42906.2
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalClassificationError = 0.54257813 * 2560; time = 0.0601s; samplesPerSecond = 42627.6
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalClassificationError = 0.54648438 * 2560; time = 0.0597s; samplesPerSecond = 42909.1
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalClassificationError = 0.53984375 * 2560; time = 0.0604s; samplesPerSecond = 42394.6
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalClassificationError = 0.52109375 * 2560; time = 0.0584s; samplesPerSecond = 43804.9
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalClassificationError = 0.52148438 * 2560; time = 0.0593s; samplesPerSecond = 43149.9
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalClassificationError = 0.51992187 * 2560; time = 0.0596s; samplesPerSecond = 42978.3
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalClassificationError = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.483777s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:17: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalErrorPrediction = 0.52470703 * 10240; time = 0.0899s; samplesPerSecond = 113873.9
|
||||
MPI Rank 2: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalErrorPrediction = 0.53974609 * 10240; time = 0.0835s; samplesPerSecond = 122592.2
|
||||
MPI Rank 2: 08/16/2016 10:00:18: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalErrorPrediction = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.179288s
|
||||
MPI Rank 2: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalClassificationError = 0.52470703 * 10240; time = 0.0899s; samplesPerSecond = 113873.9
|
||||
MPI Rank 2: 08/16/2016 10:00:18: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalClassificationError = 0.53974609 * 10240; time = 0.0835s; samplesPerSecond = 122592.2
|
||||
MPI Rank 2: 08/16/2016 10:00:18: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalClassificationError = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.179288s
|
||||
MPI Rank 2: 08/16/2016 10:00:18: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 10:00:18: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -131,7 +131,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -157,7 +157,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -234,7 +234,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -260,7 +260,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -343,7 +343,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -369,7 +369,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -465,7 +465,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -494,7 +494,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -517,7 +517,7 @@ MPI Rank 0: 08/16/2016 03:02:54: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:02:54: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:54: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:02:54: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:02:54: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -570,63 +570,63 @@ MPI Rank 0: 08/16/2016 03:02:57: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:02:57: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731197 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.0834s; samplesPerSecond = 7677.6
|
||||
MPI Rank 0: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208884 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.0883s; samplesPerSecond = 7250.5
|
||||
MPI Rank 0: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319817 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.0900s; samplesPerSecond = 7107.4
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308074 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0921s; samplesPerSecond = 6947.5
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238171 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0821s; samplesPerSecond = 7795.3
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914127 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.0791s; samplesPerSecond = 8091.8
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238447 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.0800s; samplesPerSecond = 8001.6
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740172 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.0788s; samplesPerSecond = 8121.5
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059647 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.0787s; samplesPerSecond = 8128.0
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301391 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.0875s; samplesPerSecond = 7310.2
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48831974 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.0933s; samplesPerSecond = 6859.8
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814511 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0913s; samplesPerSecond = 7010.2
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333315 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.0793s; samplesPerSecond = 8071.0
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547608 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.0767s; samplesPerSecond = 8349.4
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114572 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.0802s; samplesPerSecond = 7983.2
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450544 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.0813s; samplesPerSecond = 7876.1
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009530 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.0799s; samplesPerSecond = 8006.9
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234651 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.0793s; samplesPerSecond = 8071.0
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324351 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.0797s; samplesPerSecond = 8034.8
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050371 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.0809s; samplesPerSecond = 7915.7
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019321 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.0804s; samplesPerSecond = 7955.8
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796076 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.0809s; samplesPerSecond = 7912.9
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054678 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0805s; samplesPerSecond = 7954.5
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174440 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0838s; samplesPerSecond = 7641.2
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943276 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.0807s; samplesPerSecond = 7928.3
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070249 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.0823s; samplesPerSecond = 7774.8
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167451 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0804s; samplesPerSecond = 7961.1
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104679 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0804s; samplesPerSecond = 7959.3
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253341 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0826s; samplesPerSecond = 7749.2
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145217 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0797s; samplesPerSecond = 8032.4
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771674 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0808s; samplesPerSecond = 7918.8
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995418 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.0813s; samplesPerSecond = 7869.8
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789599 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.66361s
|
||||
MPI Rank 0: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731197 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.0834s; samplesPerSecond = 7677.6
|
||||
MPI Rank 0: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208884 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.0883s; samplesPerSecond = 7250.5
|
||||
MPI Rank 0: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319817 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.0900s; samplesPerSecond = 7107.4
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308074 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0921s; samplesPerSecond = 6947.5
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238171 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0821s; samplesPerSecond = 7795.3
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914127 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.0791s; samplesPerSecond = 8091.8
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238447 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.0800s; samplesPerSecond = 8001.6
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740172 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.0788s; samplesPerSecond = 8121.5
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059647 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.0787s; samplesPerSecond = 8128.0
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301391 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.0875s; samplesPerSecond = 7310.2
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48831974 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.0933s; samplesPerSecond = 6859.8
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814511 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0913s; samplesPerSecond = 7010.2
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333315 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.0793s; samplesPerSecond = 8071.0
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547608 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.0767s; samplesPerSecond = 8349.4
|
||||
MPI Rank 0: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114572 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.0802s; samplesPerSecond = 7983.2
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450544 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.0813s; samplesPerSecond = 7876.1
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009530 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.0799s; samplesPerSecond = 8006.9
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234651 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.0793s; samplesPerSecond = 8071.0
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324351 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.0797s; samplesPerSecond = 8034.8
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050371 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.0809s; samplesPerSecond = 7915.7
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019321 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.0804s; samplesPerSecond = 7955.8
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796076 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.0809s; samplesPerSecond = 7912.9
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054678 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0805s; samplesPerSecond = 7954.5
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174440 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0838s; samplesPerSecond = 7641.2
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943276 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.0807s; samplesPerSecond = 7928.3
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070249 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.0823s; samplesPerSecond = 7774.8
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167451 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0804s; samplesPerSecond = 7961.1
|
||||
MPI Rank 0: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104679 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0804s; samplesPerSecond = 7959.3
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253341 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0826s; samplesPerSecond = 7749.2
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145217 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0797s; samplesPerSecond = 8032.4
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771674 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0808s; samplesPerSecond = 7918.8
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995418 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.0813s; samplesPerSecond = 7869.8
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789599 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.66361s
|
||||
MPI Rank 0: 08/16/2016 03:03:00: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelNoQuantization@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.09962837 * 2560; EvalErrorPrediction = 0.56132812 * 2560; time = 0.1410s; samplesPerSecond = 18159.2
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.02412398 * 2560; EvalErrorPrediction = 0.55000000 * 2560; time = 0.1391s; samplesPerSecond = 18410.2
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.00477328 * 2560; EvalErrorPrediction = 0.54296875 * 2560; time = 0.1341s; samplesPerSecond = 19084.0
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.99184389 * 2560; EvalErrorPrediction = 0.55273438 * 2560; time = 0.1382s; samplesPerSecond = 18530.4
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.98267253 * 2560; EvalErrorPrediction = 0.54023438 * 2560; time = 0.1393s; samplesPerSecond = 18374.0
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.93130618 * 2560; EvalErrorPrediction = 0.52773437 * 2560; time = 0.1338s; samplesPerSecond = 19128.9
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91975513 * 2560; EvalErrorPrediction = 0.51718750 * 2560; time = 0.1340s; samplesPerSecond = 19098.9
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.90691429 * 2560; EvalErrorPrediction = 0.52734375 * 2560; time = 0.1373s; samplesPerSecond = 18650.3
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.98262721 * 20480; EvalErrorPrediction = 0.53994141 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.10877s
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.09962837 * 2560; EvalClassificationError = 0.56132812 * 2560; time = 0.1410s; samplesPerSecond = 18159.2
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.02412398 * 2560; EvalClassificationError = 0.55000000 * 2560; time = 0.1391s; samplesPerSecond = 18410.2
|
||||
MPI Rank 0: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.00477328 * 2560; EvalClassificationError = 0.54296875 * 2560; time = 0.1341s; samplesPerSecond = 19084.0
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.99184389 * 2560; EvalClassificationError = 0.55273438 * 2560; time = 0.1382s; samplesPerSecond = 18530.4
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.98267253 * 2560; EvalClassificationError = 0.54023438 * 2560; time = 0.1393s; samplesPerSecond = 18374.0
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.93130618 * 2560; EvalClassificationError = 0.52773437 * 2560; time = 0.1338s; samplesPerSecond = 19128.9
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91975513 * 2560; EvalClassificationError = 0.51718750 * 2560; time = 0.1340s; samplesPerSecond = 19098.9
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.90691429 * 2560; EvalClassificationError = 0.52734375 * 2560; time = 0.1373s; samplesPerSecond = 18650.3
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.98262721 * 20480; EvalClassificationError = 0.53994141 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.10877s
|
||||
MPI Rank 0: 08/16/2016 03:03:01: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelNoQuantization@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:01: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.90951136 * 10240; EvalErrorPrediction = 0.52617187 * 10240; time = 0.3803s; samplesPerSecond = 26929.2
|
||||
MPI Rank 0: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93082558 * 10240; EvalErrorPrediction = 0.54072266 * 10240; time = 0.3565s; samplesPerSecond = 28721.0
|
||||
MPI Rank 0: 08/16/2016 03:03:02: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.92016847 * 20480; EvalErrorPrediction = 0.53344727 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.746914s
|
||||
MPI Rank 0: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.90951136 * 10240; EvalClassificationError = 0.52617187 * 10240; time = 0.3803s; samplesPerSecond = 26929.2
|
||||
MPI Rank 0: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93082558 * 10240; EvalClassificationError = 0.54072266 * 10240; time = 0.3565s; samplesPerSecond = 28721.0
|
||||
MPI Rank 0: 08/16/2016 03:03:02: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.92016847 * 20480; EvalClassificationError = 0.53344727 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.746914s
|
||||
MPI Rank 0: 08/16/2016 03:03:02: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelNoQuantization@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:03:02: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -680,7 +680,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -706,7 +706,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -783,7 +783,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -809,7 +809,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -892,7 +892,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -918,7 +918,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1014,7 +1014,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1043,7 +1043,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1066,7 +1066,7 @@ MPI Rank 1: 08/16/2016 03:02:54: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:02:54: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:54: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:02:54: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:02:54: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1119,61 +1119,61 @@ MPI Rank 1: 08/16/2016 03:02:57: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:02:57: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731197 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.0834s; samplesPerSecond = 7671.1
|
||||
MPI Rank 1: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208884 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.0881s; samplesPerSecond = 7260.9
|
||||
MPI Rank 1: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319817 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.0901s; samplesPerSecond = 7105.9
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308074 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0922s; samplesPerSecond = 6939.7
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238171 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0820s; samplesPerSecond = 7807.6
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914127 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.0791s; samplesPerSecond = 8087.1
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238447 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.0800s; samplesPerSecond = 8001.9
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740172 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.0788s; samplesPerSecond = 8121.5
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059647 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.0787s; samplesPerSecond = 8136.3
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301391 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.0876s; samplesPerSecond = 7309.3
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48831974 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.0933s; samplesPerSecond = 6862.6
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814511 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0912s; samplesPerSecond = 7017.8
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333315 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.0794s; samplesPerSecond = 8065.2
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547608 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.0766s; samplesPerSecond = 8356.6
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114572 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.0802s; samplesPerSecond = 7978.7
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450544 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.0813s; samplesPerSecond = 7873.7
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009530 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.0799s; samplesPerSecond = 8006.7
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234651 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.0793s; samplesPerSecond = 8071.2
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324351 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.0796s; samplesPerSecond = 8035.6
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050371 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.0809s; samplesPerSecond = 7914.7
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019321 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.0805s; samplesPerSecond = 7947.3
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796076 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.0809s; samplesPerSecond = 7906.9
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054678 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0805s; samplesPerSecond = 7951.8
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174440 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0838s; samplesPerSecond = 7639.1
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943276 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.0808s; samplesPerSecond = 7924.7
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070249 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.0823s; samplesPerSecond = 7776.0
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167451 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0804s; samplesPerSecond = 7963.1
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104679 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0804s; samplesPerSecond = 7955.3
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253341 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0825s; samplesPerSecond = 7753.2
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145217 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0797s; samplesPerSecond = 8032.7
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771674 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0808s; samplesPerSecond = 7923.1
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995418 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.0813s; samplesPerSecond = 7867.7
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789599 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.66361s
|
||||
MPI Rank 1: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731197 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.0834s; samplesPerSecond = 7671.1
|
||||
MPI Rank 1: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208884 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.0881s; samplesPerSecond = 7260.9
|
||||
MPI Rank 1: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319817 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.0901s; samplesPerSecond = 7105.9
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308074 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0922s; samplesPerSecond = 6939.7
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238171 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0820s; samplesPerSecond = 7807.6
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914127 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.0791s; samplesPerSecond = 8087.1
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238447 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.0800s; samplesPerSecond = 8001.9
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740172 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.0788s; samplesPerSecond = 8121.5
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059647 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.0787s; samplesPerSecond = 8136.3
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301391 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.0876s; samplesPerSecond = 7309.3
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48831974 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.0933s; samplesPerSecond = 6862.6
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814511 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0912s; samplesPerSecond = 7017.8
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333315 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.0794s; samplesPerSecond = 8065.2
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547608 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.0766s; samplesPerSecond = 8356.6
|
||||
MPI Rank 1: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114572 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.0802s; samplesPerSecond = 7978.7
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450544 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.0813s; samplesPerSecond = 7873.7
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009530 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.0799s; samplesPerSecond = 8006.7
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234651 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.0793s; samplesPerSecond = 8071.2
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324351 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.0796s; samplesPerSecond = 8035.6
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050371 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.0809s; samplesPerSecond = 7914.7
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019321 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.0805s; samplesPerSecond = 7947.3
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796076 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.0809s; samplesPerSecond = 7906.9
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054678 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0805s; samplesPerSecond = 7951.8
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174440 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0838s; samplesPerSecond = 7639.1
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943276 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.0808s; samplesPerSecond = 7924.7
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070249 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.0823s; samplesPerSecond = 7776.0
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167451 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0804s; samplesPerSecond = 7963.1
|
||||
MPI Rank 1: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104679 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0804s; samplesPerSecond = 7955.3
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253341 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0825s; samplesPerSecond = 7753.2
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145217 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0797s; samplesPerSecond = 8032.7
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771674 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0808s; samplesPerSecond = 7923.1
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995418 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.0813s; samplesPerSecond = 7867.7
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789599 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.66361s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.09962837 * 2560; EvalErrorPrediction = 0.56132812 * 2560; time = 0.1412s; samplesPerSecond = 18128.6
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.02412398 * 2560; EvalErrorPrediction = 0.55000000 * 2560; time = 0.1389s; samplesPerSecond = 18430.0
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.00477328 * 2560; EvalErrorPrediction = 0.54296875 * 2560; time = 0.1342s; samplesPerSecond = 19081.7
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.99184389 * 2560; EvalErrorPrediction = 0.55273438 * 2560; time = 0.1382s; samplesPerSecond = 18523.2
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.98267253 * 2560; EvalErrorPrediction = 0.54023438 * 2560; time = 0.1392s; samplesPerSecond = 18389.4
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.93130618 * 2560; EvalErrorPrediction = 0.52773437 * 2560; time = 0.1339s; samplesPerSecond = 19114.3
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91975513 * 2560; EvalErrorPrediction = 0.51718750 * 2560; time = 0.1342s; samplesPerSecond = 19070.5
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.90691429 * 2560; EvalErrorPrediction = 0.52734375 * 2560; time = 0.1371s; samplesPerSecond = 18668.7
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.98262721 * 20480; EvalErrorPrediction = 0.53994141 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.10877s
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.09962837 * 2560; EvalClassificationError = 0.56132812 * 2560; time = 0.1412s; samplesPerSecond = 18128.6
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.02412398 * 2560; EvalClassificationError = 0.55000000 * 2560; time = 0.1389s; samplesPerSecond = 18430.0
|
||||
MPI Rank 1: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.00477328 * 2560; EvalClassificationError = 0.54296875 * 2560; time = 0.1342s; samplesPerSecond = 19081.7
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.99184389 * 2560; EvalClassificationError = 0.55273438 * 2560; time = 0.1382s; samplesPerSecond = 18523.2
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.98267253 * 2560; EvalClassificationError = 0.54023438 * 2560; time = 0.1392s; samplesPerSecond = 18389.4
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.93130618 * 2560; EvalClassificationError = 0.52773437 * 2560; time = 0.1339s; samplesPerSecond = 19114.3
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91975513 * 2560; EvalClassificationError = 0.51718750 * 2560; time = 0.1342s; samplesPerSecond = 19070.5
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.90691429 * 2560; EvalClassificationError = 0.52734375 * 2560; time = 0.1371s; samplesPerSecond = 18668.7
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.98262721 * 20480; EvalClassificationError = 0.53994141 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.10877s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:01: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.90951136 * 10240; EvalErrorPrediction = 0.52617187 * 10240; time = 0.3812s; samplesPerSecond = 26864.6
|
||||
MPI Rank 1: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93082558 * 10240; EvalErrorPrediction = 0.54072266 * 10240; time = 0.3564s; samplesPerSecond = 28728.9
|
||||
MPI Rank 1: 08/16/2016 03:03:02: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.92016847 * 20480; EvalErrorPrediction = 0.53344727 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.746915s
|
||||
MPI Rank 1: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.90951136 * 10240; EvalClassificationError = 0.52617187 * 10240; time = 0.3812s; samplesPerSecond = 26864.6
|
||||
MPI Rank 1: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93082558 * 10240; EvalClassificationError = 0.54072266 * 10240; time = 0.3564s; samplesPerSecond = 28728.9
|
||||
MPI Rank 1: 08/16/2016 03:03:02: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.92016847 * 20480; EvalClassificationError = 0.53344727 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.746915s
|
||||
MPI Rank 1: 08/16/2016 03:03:02: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:02: Action "train" complete.
|
||||
|
|
@ -1226,7 +1226,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1252,7 +1252,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1329,7 +1329,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1355,7 +1355,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1438,7 +1438,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1464,7 +1464,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1560,7 +1560,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1589,7 +1589,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1612,7 +1612,7 @@ MPI Rank 2: 08/16/2016 03:02:55: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 03:02:55: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:02:55: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 03:02:55: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 03:02:55: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1665,61 +1665,61 @@ MPI Rank 2: 08/16/2016 03:02:57: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:02:57: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731197 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.0829s; samplesPerSecond = 7722.7
|
||||
MPI Rank 2: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208884 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.0881s; samplesPerSecond = 7266.9
|
||||
MPI Rank 2: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319817 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.0901s; samplesPerSecond = 7102.3
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308074 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0921s; samplesPerSecond = 6947.8
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238171 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0820s; samplesPerSecond = 7803.8
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914127 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.0791s; samplesPerSecond = 8095.6
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238447 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.0799s; samplesPerSecond = 8014.1
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740172 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.0788s; samplesPerSecond = 8117.4
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059647 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.0787s; samplesPerSecond = 8129.0
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301391 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.0875s; samplesPerSecond = 7312.1
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48831974 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.0932s; samplesPerSecond = 6867.2
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814511 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0912s; samplesPerSecond = 7019.5
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333315 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.0794s; samplesPerSecond = 8062.9
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547608 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.0765s; samplesPerSecond = 8362.4
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114572 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.0802s; samplesPerSecond = 7980.1
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450544 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.0812s; samplesPerSecond = 7877.5
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009530 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.0799s; samplesPerSecond = 8007.2
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234651 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.0793s; samplesPerSecond = 8074.3
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324351 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.0796s; samplesPerSecond = 8040.0
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050371 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.0808s; samplesPerSecond = 7921.2
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019321 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.0805s; samplesPerSecond = 7947.9
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796076 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.0809s; samplesPerSecond = 7913.3
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054678 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0804s; samplesPerSecond = 7957.1
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174440 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0837s; samplesPerSecond = 7642.4
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943276 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.0807s; samplesPerSecond = 7928.1
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070249 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.0823s; samplesPerSecond = 7778.5
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167451 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0803s; samplesPerSecond = 7965.4
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104679 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0804s; samplesPerSecond = 7962.3
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253341 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0824s; samplesPerSecond = 7771.2
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145217 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0798s; samplesPerSecond = 8025.0
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771674 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.0807s; samplesPerSecond = 7928.7
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995418 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.0814s; samplesPerSecond = 7863.9
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789599 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.66356s
|
||||
MPI Rank 2: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731197 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.0829s; samplesPerSecond = 7722.7
|
||||
MPI Rank 2: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208884 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.0881s; samplesPerSecond = 7266.9
|
||||
MPI Rank 2: 08/16/2016 03:02:57: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319817 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.0901s; samplesPerSecond = 7102.3
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308074 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0921s; samplesPerSecond = 6947.8
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238171 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0820s; samplesPerSecond = 7803.8
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914127 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.0791s; samplesPerSecond = 8095.6
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238447 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.0799s; samplesPerSecond = 8014.1
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740172 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.0788s; samplesPerSecond = 8117.4
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059647 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.0787s; samplesPerSecond = 8129.0
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301391 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.0875s; samplesPerSecond = 7312.1
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48831974 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.0932s; samplesPerSecond = 6867.2
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814511 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0912s; samplesPerSecond = 7019.5
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333315 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.0794s; samplesPerSecond = 8062.9
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547608 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.0765s; samplesPerSecond = 8362.4
|
||||
MPI Rank 2: 08/16/2016 03:02:58: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114572 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.0802s; samplesPerSecond = 7980.1
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450544 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.0812s; samplesPerSecond = 7877.5
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009530 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.0799s; samplesPerSecond = 8007.2
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234651 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.0793s; samplesPerSecond = 8074.3
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324351 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.0796s; samplesPerSecond = 8040.0
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050371 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.0808s; samplesPerSecond = 7921.2
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019321 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.0805s; samplesPerSecond = 7947.9
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796076 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.0809s; samplesPerSecond = 7913.3
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054678 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0804s; samplesPerSecond = 7957.1
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174440 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0837s; samplesPerSecond = 7642.4
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943276 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.0807s; samplesPerSecond = 7928.1
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070249 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.0823s; samplesPerSecond = 7778.5
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167451 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0803s; samplesPerSecond = 7965.4
|
||||
MPI Rank 2: 08/16/2016 03:02:59: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104679 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0804s; samplesPerSecond = 7962.3
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253341 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0824s; samplesPerSecond = 7771.2
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145217 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0798s; samplesPerSecond = 8025.0
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771674 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.0807s; samplesPerSecond = 7928.7
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995418 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.0814s; samplesPerSecond = 7863.9
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 3.00789599 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.66356s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.09962837 * 2560; EvalErrorPrediction = 0.56132812 * 2560; time = 0.1413s; samplesPerSecond = 18119.0
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.02412398 * 2560; EvalErrorPrediction = 0.55000000 * 2560; time = 0.1389s; samplesPerSecond = 18436.9
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.00477328 * 2560; EvalErrorPrediction = 0.54296875 * 2560; time = 0.1341s; samplesPerSecond = 19089.9
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.99184389 * 2560; EvalErrorPrediction = 0.55273438 * 2560; time = 0.1382s; samplesPerSecond = 18530.4
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.98267253 * 2560; EvalErrorPrediction = 0.54023438 * 2560; time = 0.1393s; samplesPerSecond = 18373.4
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.93130618 * 2560; EvalErrorPrediction = 0.52773437 * 2560; time = 0.1338s; samplesPerSecond = 19131.2
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91975513 * 2560; EvalErrorPrediction = 0.51718750 * 2560; time = 0.1341s; samplesPerSecond = 19088.4
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.90691429 * 2560; EvalErrorPrediction = 0.52734375 * 2560; time = 0.1372s; samplesPerSecond = 18663.5
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.98262721 * 20480; EvalErrorPrediction = 0.53994141 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.10871s
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.09962837 * 2560; EvalClassificationError = 0.56132812 * 2560; time = 0.1413s; samplesPerSecond = 18119.0
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.02412398 * 2560; EvalClassificationError = 0.55000000 * 2560; time = 0.1389s; samplesPerSecond = 18436.9
|
||||
MPI Rank 2: 08/16/2016 03:03:00: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.00477328 * 2560; EvalClassificationError = 0.54296875 * 2560; time = 0.1341s; samplesPerSecond = 19089.9
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.99184389 * 2560; EvalClassificationError = 0.55273438 * 2560; time = 0.1382s; samplesPerSecond = 18530.4
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.98267253 * 2560; EvalClassificationError = 0.54023438 * 2560; time = 0.1393s; samplesPerSecond = 18373.4
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.93130618 * 2560; EvalClassificationError = 0.52773437 * 2560; time = 0.1338s; samplesPerSecond = 19131.2
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91975513 * 2560; EvalClassificationError = 0.51718750 * 2560; time = 0.1341s; samplesPerSecond = 19088.4
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.90691429 * 2560; EvalClassificationError = 0.52734375 * 2560; time = 0.1372s; samplesPerSecond = 18663.5
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.98262721 * 20480; EvalClassificationError = 0.53994141 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.10871s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:01: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.90951136 * 10240; EvalErrorPrediction = 0.52617187 * 10240; time = 0.3813s; samplesPerSecond = 26856.3
|
||||
MPI Rank 2: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93082558 * 10240; EvalErrorPrediction = 0.54072266 * 10240; time = 0.3565s; samplesPerSecond = 28725.1
|
||||
MPI Rank 2: 08/16/2016 03:03:02: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.92016847 * 20480; EvalErrorPrediction = 0.53344727 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.746861s
|
||||
MPI Rank 2: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.90951136 * 10240; EvalClassificationError = 0.52617187 * 10240; time = 0.3813s; samplesPerSecond = 26856.3
|
||||
MPI Rank 2: 08/16/2016 03:03:02: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93082558 * 10240; EvalClassificationError = 0.54072266 * 10240; time = 0.3565s; samplesPerSecond = 28725.1
|
||||
MPI Rank 2: 08/16/2016 03:03:02: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.92016847 * 20480; EvalClassificationError = 0.53344727 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.746861s
|
||||
MPI Rank 2: 08/16/2016 03:03:02: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:02: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -131,7 +131,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -157,7 +157,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -234,7 +234,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -260,7 +260,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -343,7 +343,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -369,7 +369,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -466,7 +466,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -495,7 +495,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -518,7 +518,7 @@ MPI Rank 0: 08/16/2016 03:03:12: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:03:12: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:12: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:03:12: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:03:12: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -571,63 +571,63 @@ MPI Rank 0: 08/16/2016 03:03:17: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalErrorPrediction = 0.92031250 * 640; time = 0.1000s; samplesPerSecond = 6400.3
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.0738s; samplesPerSecond = 8674.2
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalErrorPrediction = 0.87187500 * 640; time = 0.0762s; samplesPerSecond = 8396.9
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0776s; samplesPerSecond = 8246.0
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0708s; samplesPerSecond = 9036.0
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.0704s; samplesPerSecond = 9089.1
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0743s; samplesPerSecond = 8611.9
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalErrorPrediction = 0.82968750 * 640; time = 0.0707s; samplesPerSecond = 9046.8
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalErrorPrediction = 0.81093750 * 640; time = 0.0700s; samplesPerSecond = 9141.9
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.0692s; samplesPerSecond = 9245.5
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0771s; samplesPerSecond = 8300.8
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalErrorPrediction = 0.75625000 * 640; time = 0.0736s; samplesPerSecond = 8691.4
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalErrorPrediction = 0.75312500 * 640; time = 0.0707s; samplesPerSecond = 9054.4
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0705s; samplesPerSecond = 9075.3
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalErrorPrediction = 0.70000000 * 640; time = 0.0723s; samplesPerSecond = 8848.7
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalErrorPrediction = 0.74218750 * 640; time = 0.0744s; samplesPerSecond = 8607.6
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalErrorPrediction = 0.69375000 * 640; time = 0.0714s; samplesPerSecond = 8962.2
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0714s; samplesPerSecond = 8967.5
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalErrorPrediction = 0.69062500 * 640; time = 0.0729s; samplesPerSecond = 8776.4
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalErrorPrediction = 0.68593750 * 640; time = 0.0738s; samplesPerSecond = 8674.9
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalErrorPrediction = 0.64375000 * 640; time = 0.0730s; samplesPerSecond = 8770.7
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0724s; samplesPerSecond = 8838.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0723s; samplesPerSecond = 8847.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.0724s; samplesPerSecond = 8841.9
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalErrorPrediction = 0.63125000 * 640; time = 0.0725s; samplesPerSecond = 8823.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalErrorPrediction = 0.61562500 * 640; time = 0.0722s; samplesPerSecond = 8869.2
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0738s; samplesPerSecond = 8669.2
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0739s; samplesPerSecond = 8657.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0731s; samplesPerSecond = 8755.4
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalErrorPrediction = 0.60781250 * 640; time = 0.0717s; samplesPerSecond = 8930.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0718s; samplesPerSecond = 8919.1
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalErrorPrediction = 0.58593750 * 640; time = 0.0727s; samplesPerSecond = 8807.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalErrorPrediction = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.37732s
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalClassificationError = 0.92031250 * 640; time = 0.1000s; samplesPerSecond = 6400.3
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.0738s; samplesPerSecond = 8674.2
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalClassificationError = 0.87187500 * 640; time = 0.0762s; samplesPerSecond = 8396.9
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0776s; samplesPerSecond = 8246.0
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0708s; samplesPerSecond = 9036.0
|
||||
MPI Rank 0: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.0704s; samplesPerSecond = 9089.1
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0743s; samplesPerSecond = 8611.9
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalClassificationError = 0.82968750 * 640; time = 0.0707s; samplesPerSecond = 9046.8
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalClassificationError = 0.81093750 * 640; time = 0.0700s; samplesPerSecond = 9141.9
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.0692s; samplesPerSecond = 9245.5
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0771s; samplesPerSecond = 8300.8
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalClassificationError = 0.75625000 * 640; time = 0.0736s; samplesPerSecond = 8691.4
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalClassificationError = 0.75312500 * 640; time = 0.0707s; samplesPerSecond = 9054.4
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0705s; samplesPerSecond = 9075.3
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalClassificationError = 0.70000000 * 640; time = 0.0723s; samplesPerSecond = 8848.7
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalClassificationError = 0.74218750 * 640; time = 0.0744s; samplesPerSecond = 8607.6
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalClassificationError = 0.69375000 * 640; time = 0.0714s; samplesPerSecond = 8962.2
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0714s; samplesPerSecond = 8967.5
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalClassificationError = 0.69062500 * 640; time = 0.0729s; samplesPerSecond = 8776.4
|
||||
MPI Rank 0: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalClassificationError = 0.68593750 * 640; time = 0.0738s; samplesPerSecond = 8674.9
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalClassificationError = 0.64375000 * 640; time = 0.0730s; samplesPerSecond = 8770.7
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0724s; samplesPerSecond = 8838.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0723s; samplesPerSecond = 8847.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.0724s; samplesPerSecond = 8841.9
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalClassificationError = 0.63125000 * 640; time = 0.0725s; samplesPerSecond = 8823.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalClassificationError = 0.61562500 * 640; time = 0.0722s; samplesPerSecond = 8869.2
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0738s; samplesPerSecond = 8669.2
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0739s; samplesPerSecond = 8657.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0731s; samplesPerSecond = 8755.4
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalClassificationError = 0.60781250 * 640; time = 0.0717s; samplesPerSecond = 8930.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0718s; samplesPerSecond = 8919.1
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalClassificationError = 0.58593750 * 640; time = 0.0727s; samplesPerSecond = 8807.8
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalClassificationError = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.37732s
|
||||
MPI Rank 0: 08/16/2016 03:03:19: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelNoQuantization@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:19: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalErrorPrediction = 0.56367188 * 2560; time = 0.1045s; samplesPerSecond = 24493.4
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalErrorPrediction = 0.54218750 * 2560; time = 0.0862s; samplesPerSecond = 29707.7
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalErrorPrediction = 0.54257813 * 2560; time = 0.0768s; samplesPerSecond = 33337.7
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalErrorPrediction = 0.54648438 * 2560; time = 0.0775s; samplesPerSecond = 33016.5
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalErrorPrediction = 0.53984375 * 2560; time = 0.0762s; samplesPerSecond = 33575.1
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalErrorPrediction = 0.52109375 * 2560; time = 0.0803s; samplesPerSecond = 31884.8
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalErrorPrediction = 0.52148438 * 2560; time = 0.0779s; samplesPerSecond = 32853.8
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalErrorPrediction = 0.51992187 * 2560; time = 0.0796s; samplesPerSecond = 32165.2
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalErrorPrediction = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.671006s
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalClassificationError = 0.56367188 * 2560; time = 0.1045s; samplesPerSecond = 24493.4
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalClassificationError = 0.54218750 * 2560; time = 0.0862s; samplesPerSecond = 29707.7
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalClassificationError = 0.54257813 * 2560; time = 0.0768s; samplesPerSecond = 33337.7
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalClassificationError = 0.54648438 * 2560; time = 0.0775s; samplesPerSecond = 33016.5
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalClassificationError = 0.53984375 * 2560; time = 0.0762s; samplesPerSecond = 33575.1
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalClassificationError = 0.52109375 * 2560; time = 0.0803s; samplesPerSecond = 31884.8
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalClassificationError = 0.52148438 * 2560; time = 0.0779s; samplesPerSecond = 32853.8
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalClassificationError = 0.51992187 * 2560; time = 0.0796s; samplesPerSecond = 32165.2
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalClassificationError = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.671006s
|
||||
MPI Rank 0: 08/16/2016 03:03:20: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelNoQuantization@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 0: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 0 of 3, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Starting minibatch loop, DataParallelSGD training (MyRank = 0, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalErrorPrediction = 0.52470703 * 10240; time = 0.1518s; samplesPerSecond = 67460.3
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalErrorPrediction = 0.53974609 * 10240; time = 0.1251s; samplesPerSecond = 81832.3
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalErrorPrediction = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.287003s
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalClassificationError = 0.52470703 * 10240; time = 0.1518s; samplesPerSecond = 67460.3
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalClassificationError = 0.53974609 * 10240; time = 0.1251s; samplesPerSecond = 81832.3
|
||||
MPI Rank 0: 08/16/2016 03:03:20: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalClassificationError = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.287003s
|
||||
MPI Rank 0: 08/16/2016 03:03:21: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030048.672180\Speech\DNN_ParallelNoQuantization@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:03:21: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -681,7 +681,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -707,7 +707,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -784,7 +784,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -810,7 +810,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -893,7 +893,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -919,7 +919,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1016,7 +1016,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1045,7 +1045,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1068,7 +1068,7 @@ MPI Rank 1: 08/16/2016 03:03:13: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:03:13: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:13: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:03:13: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:03:13: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1121,61 +1121,61 @@ MPI Rank 1: 08/16/2016 03:03:17: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalErrorPrediction = 0.92031250 * 640; time = 0.0996s; samplesPerSecond = 6427.7
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.0742s; samplesPerSecond = 8629.6
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalErrorPrediction = 0.87187500 * 640; time = 0.0762s; samplesPerSecond = 8393.6
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0772s; samplesPerSecond = 8286.7
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0708s; samplesPerSecond = 9040.3
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.0708s; samplesPerSecond = 9036.1
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0740s; samplesPerSecond = 8644.4
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalErrorPrediction = 0.82968750 * 640; time = 0.0711s; samplesPerSecond = 9006.3
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalErrorPrediction = 0.81093750 * 640; time = 0.0696s; samplesPerSecond = 9193.7
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.0696s; samplesPerSecond = 9191.0
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0771s; samplesPerSecond = 8298.3
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalErrorPrediction = 0.75625000 * 640; time = 0.0733s; samplesPerSecond = 8736.6
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalErrorPrediction = 0.75312500 * 640; time = 0.0711s; samplesPerSecond = 9006.1
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0706s; samplesPerSecond = 9069.1
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalErrorPrediction = 0.70000000 * 640; time = 0.0724s; samplesPerSecond = 8842.7
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalErrorPrediction = 0.74218750 * 640; time = 0.0739s; samplesPerSecond = 8654.6
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalErrorPrediction = 0.69375000 * 640; time = 0.0713s; samplesPerSecond = 8971.5
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0718s; samplesPerSecond = 8909.1
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalErrorPrediction = 0.69062500 * 640; time = 0.0729s; samplesPerSecond = 8774.1
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalErrorPrediction = 0.68593750 * 640; time = 0.0738s; samplesPerSecond = 8671.7
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalErrorPrediction = 0.64375000 * 640; time = 0.0726s; samplesPerSecond = 8820.2
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0724s; samplesPerSecond = 8840.9
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0723s; samplesPerSecond = 8851.0
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.0728s; samplesPerSecond = 8792.1
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalErrorPrediction = 0.63125000 * 640; time = 0.0726s; samplesPerSecond = 8820.9
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalErrorPrediction = 0.61562500 * 640; time = 0.0718s; samplesPerSecond = 8918.6
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0738s; samplesPerSecond = 8671.0
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0743s; samplesPerSecond = 8608.4
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0727s; samplesPerSecond = 8801.6
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalErrorPrediction = 0.60781250 * 640; time = 0.0720s; samplesPerSecond = 8883.0
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0718s; samplesPerSecond = 8914.4
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalErrorPrediction = 0.58593750 * 640; time = 0.0723s; samplesPerSecond = 8857.2
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalErrorPrediction = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.37714s
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalClassificationError = 0.92031250 * 640; time = 0.0996s; samplesPerSecond = 6427.7
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.0742s; samplesPerSecond = 8629.6
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalClassificationError = 0.87187500 * 640; time = 0.0762s; samplesPerSecond = 8393.6
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0772s; samplesPerSecond = 8286.7
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0708s; samplesPerSecond = 9040.3
|
||||
MPI Rank 1: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.0708s; samplesPerSecond = 9036.1
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0740s; samplesPerSecond = 8644.4
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalClassificationError = 0.82968750 * 640; time = 0.0711s; samplesPerSecond = 9006.3
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalClassificationError = 0.81093750 * 640; time = 0.0696s; samplesPerSecond = 9193.7
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.0696s; samplesPerSecond = 9191.0
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0771s; samplesPerSecond = 8298.3
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalClassificationError = 0.75625000 * 640; time = 0.0733s; samplesPerSecond = 8736.6
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalClassificationError = 0.75312500 * 640; time = 0.0711s; samplesPerSecond = 9006.1
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0706s; samplesPerSecond = 9069.1
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalClassificationError = 0.70000000 * 640; time = 0.0724s; samplesPerSecond = 8842.7
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalClassificationError = 0.74218750 * 640; time = 0.0739s; samplesPerSecond = 8654.6
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalClassificationError = 0.69375000 * 640; time = 0.0713s; samplesPerSecond = 8971.5
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0718s; samplesPerSecond = 8909.1
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalClassificationError = 0.69062500 * 640; time = 0.0729s; samplesPerSecond = 8774.1
|
||||
MPI Rank 1: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalClassificationError = 0.68593750 * 640; time = 0.0738s; samplesPerSecond = 8671.7
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalClassificationError = 0.64375000 * 640; time = 0.0726s; samplesPerSecond = 8820.2
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0724s; samplesPerSecond = 8840.9
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0723s; samplesPerSecond = 8851.0
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.0728s; samplesPerSecond = 8792.1
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalClassificationError = 0.63125000 * 640; time = 0.0726s; samplesPerSecond = 8820.9
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalClassificationError = 0.61562500 * 640; time = 0.0718s; samplesPerSecond = 8918.6
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0738s; samplesPerSecond = 8671.0
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0743s; samplesPerSecond = 8608.4
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0727s; samplesPerSecond = 8801.6
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalClassificationError = 0.60781250 * 640; time = 0.0720s; samplesPerSecond = 8883.0
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0718s; samplesPerSecond = 8914.4
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalClassificationError = 0.58593750 * 640; time = 0.0723s; samplesPerSecond = 8857.2
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalClassificationError = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.37714s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:19: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalErrorPrediction = 0.56367188 * 2560; time = 0.1047s; samplesPerSecond = 24439.6
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalErrorPrediction = 0.54218750 * 2560; time = 0.0858s; samplesPerSecond = 29832.0
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalErrorPrediction = 0.54257813 * 2560; time = 0.0767s; samplesPerSecond = 33361.1
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalErrorPrediction = 0.54648438 * 2560; time = 0.0775s; samplesPerSecond = 33029.7
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalErrorPrediction = 0.53984375 * 2560; time = 0.0766s; samplesPerSecond = 33400.3
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalErrorPrediction = 0.52109375 * 2560; time = 0.0804s; samplesPerSecond = 31829.3
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalErrorPrediction = 0.52148438 * 2560; time = 0.0779s; samplesPerSecond = 32851.7
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalErrorPrediction = 0.51992187 * 2560; time = 0.0796s; samplesPerSecond = 32153.9
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalErrorPrediction = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.671177s
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalClassificationError = 0.56367188 * 2560; time = 0.1047s; samplesPerSecond = 24439.6
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalClassificationError = 0.54218750 * 2560; time = 0.0858s; samplesPerSecond = 29832.0
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalClassificationError = 0.54257813 * 2560; time = 0.0767s; samplesPerSecond = 33361.1
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalClassificationError = 0.54648438 * 2560; time = 0.0775s; samplesPerSecond = 33029.7
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalClassificationError = 0.53984375 * 2560; time = 0.0766s; samplesPerSecond = 33400.3
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalClassificationError = 0.52109375 * 2560; time = 0.0804s; samplesPerSecond = 31829.3
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalClassificationError = 0.52148438 * 2560; time = 0.0779s; samplesPerSecond = 32851.7
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalClassificationError = 0.51992187 * 2560; time = 0.0796s; samplesPerSecond = 32153.9
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalClassificationError = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.671177s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Starting minibatch loop, DataParallelSGD training (MyRank = 1, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalErrorPrediction = 0.52470703 * 10240; time = 0.1520s; samplesPerSecond = 67362.2
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalErrorPrediction = 0.53974609 * 10240; time = 0.1252s; samplesPerSecond = 81779.3
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalErrorPrediction = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.287153s
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalClassificationError = 0.52470703 * 10240; time = 0.1520s; samplesPerSecond = 67362.2
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalClassificationError = 0.53974609 * 10240; time = 0.1252s; samplesPerSecond = 81779.3
|
||||
MPI Rank 1: 08/16/2016 03:03:20: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalClassificationError = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.287153s
|
||||
MPI Rank 1: 08/16/2016 03:03:21: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:21: Action "train" complete.
|
||||
|
|
@ -1228,7 +1228,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1254,7 +1254,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1331,7 +1331,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1357,7 +1357,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1440,7 +1440,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1466,7 +1466,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1563,7 +1563,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1592,7 +1592,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1615,7 +1615,7 @@ MPI Rank 2: 08/16/2016 03:03:14: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 03:03:14: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:14: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 03:03:14: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 03:03:14: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1668,61 +1668,61 @@ MPI Rank 2: 08/16/2016 03:03:17: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalErrorPrediction = 0.92031250 * 640; time = 0.0986s; samplesPerSecond = 6490.4
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.0740s; samplesPerSecond = 8642.9
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalErrorPrediction = 0.87187500 * 640; time = 0.0763s; samplesPerSecond = 8383.7
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0773s; samplesPerSecond = 8274.6
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.0709s; samplesPerSecond = 9030.0
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.0707s; samplesPerSecond = 9054.3
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.0750s; samplesPerSecond = 8535.4
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalErrorPrediction = 0.82968750 * 640; time = 0.0696s; samplesPerSecond = 9189.3
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalErrorPrediction = 0.81093750 * 640; time = 0.0699s; samplesPerSecond = 9159.5
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.0695s; samplesPerSecond = 9210.8
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0772s; samplesPerSecond = 8290.7
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalErrorPrediction = 0.75625000 * 640; time = 0.0735s; samplesPerSecond = 8702.5
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalErrorPrediction = 0.75312500 * 640; time = 0.0709s; samplesPerSecond = 9024.5
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0706s; samplesPerSecond = 9063.0
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalErrorPrediction = 0.70000000 * 640; time = 0.0724s; samplesPerSecond = 8834.8
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalErrorPrediction = 0.74218750 * 640; time = 0.0742s; samplesPerSecond = 8623.6
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalErrorPrediction = 0.69375000 * 640; time = 0.0714s; samplesPerSecond = 8959.1
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0717s; samplesPerSecond = 8926.3
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalErrorPrediction = 0.69062500 * 640; time = 0.0730s; samplesPerSecond = 8762.6
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalErrorPrediction = 0.68593750 * 640; time = 0.0739s; samplesPerSecond = 8664.5
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalErrorPrediction = 0.64375000 * 640; time = 0.0728s; samplesPerSecond = 8787.5
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalErrorPrediction = 0.63750000 * 640; time = 0.0725s; samplesPerSecond = 8831.4
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0724s; samplesPerSecond = 8839.2
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.0727s; samplesPerSecond = 8804.8
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalErrorPrediction = 0.63125000 * 640; time = 0.0726s; samplesPerSecond = 8812.3
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalErrorPrediction = 0.61562500 * 640; time = 0.0720s; samplesPerSecond = 8882.8
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0739s; samplesPerSecond = 8661.9
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0742s; samplesPerSecond = 8628.4
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0730s; samplesPerSecond = 8770.6
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalErrorPrediction = 0.60781250 * 640; time = 0.0720s; samplesPerSecond = 8893.0
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalErrorPrediction = 0.57812500 * 640; time = 0.0718s; samplesPerSecond = 8912.4
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalErrorPrediction = 0.58593750 * 640; time = 0.0725s; samplesPerSecond = 8821.9
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalErrorPrediction = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.37701s
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.53638629 * 640; EvalClassificationError = 0.92031250 * 640; time = 0.0986s; samplesPerSecond = 6490.4
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.32517790 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.0740s; samplesPerSecond = 8642.9
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.98246287 * 640; EvalClassificationError = 0.87187500 * 640; time = 0.0763s; samplesPerSecond = 8383.7
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73673605 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0773s; samplesPerSecond = 8274.6
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.84021875 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.0709s; samplesPerSecond = 9030.0
|
||||
MPI Rank 2: 08/16/2016 03:03:17: Epoch[ 1 of 3]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69831380 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.0707s; samplesPerSecond = 9054.3
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.39593109 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.0750s; samplesPerSecond = 8535.4
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.49749692 * 640; EvalClassificationError = 0.82968750 * 640; time = 0.0696s; samplesPerSecond = 9189.3
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.47295704 * 640; EvalClassificationError = 0.81093750 * 640; time = 0.0699s; samplesPerSecond = 9159.5
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.36483702 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.0695s; samplesPerSecond = 9210.8
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.46790690 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0772s; samplesPerSecond = 8290.7
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.22104746 * 640; EvalClassificationError = 0.75625000 * 640; time = 0.0735s; samplesPerSecond = 8702.5
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.12504338 * 640; EvalClassificationError = 0.75312500 * 640; time = 0.0709s; samplesPerSecond = 9024.5
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 2.99508079 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0706s; samplesPerSecond = 9063.0
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.89602891 * 640; EvalClassificationError = 0.70000000 * 640; time = 0.0724s; samplesPerSecond = 8834.8
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.04740224 * 640; EvalClassificationError = 0.74218750 * 640; time = 0.0742s; samplesPerSecond = 8623.6
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.75064614 * 640; EvalClassificationError = 0.69375000 * 640; time = 0.0714s; samplesPerSecond = 8959.1
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.65538367 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0717s; samplesPerSecond = 8926.3
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.74816077 * 640; EvalClassificationError = 0.69062500 * 640; time = 0.0730s; samplesPerSecond = 8762.6
|
||||
MPI Rank 2: 08/16/2016 03:03:18: Epoch[ 1 of 3]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.68736711 * 640; EvalClassificationError = 0.68593750 * 640; time = 0.0739s; samplesPerSecond = 8664.5
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.53268725 * 640; EvalClassificationError = 0.64375000 * 640; time = 0.0728s; samplesPerSecond = 8787.5
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.53923326 * 640; EvalClassificationError = 0.63750000 * 640; time = 0.0725s; samplesPerSecond = 8831.4
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.48909469 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0724s; samplesPerSecond = 8839.2
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.50033043 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.0727s; samplesPerSecond = 8804.8
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.43569649 * 640; EvalClassificationError = 0.63125000 * 640; time = 0.0726s; samplesPerSecond = 8812.3
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.34293081 * 640; EvalClassificationError = 0.61562500 * 640; time = 0.0720s; samplesPerSecond = 8882.8
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.20428051 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0739s; samplesPerSecond = 8661.9
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.46886810 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0742s; samplesPerSecond = 8628.4
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.22066712 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0730s; samplesPerSecond = 8770.6
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.21784272 * 640; EvalClassificationError = 0.60781250 * 640; time = 0.0720s; samplesPerSecond = 8893.0
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.20442216 * 640; EvalClassificationError = 0.57812500 * 640; time = 0.0718s; samplesPerSecond = 8912.4
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Epoch[ 1 of 3]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.18215667 * 640; EvalClassificationError = 0.58593750 * 640; time = 0.0725s; samplesPerSecond = 8821.9
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Finished Epoch[ 1 of 3]: [Training] CrossEntropyWithSoftmax = 2.99321235 * 20480; EvalClassificationError = 0.72216797 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=2.37701s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:19: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalErrorPrediction = 0.56367188 * 2560; time = 0.1034s; samplesPerSecond = 24758.7
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalErrorPrediction = 0.54218750 * 2560; time = 0.0861s; samplesPerSecond = 29745.7
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalErrorPrediction = 0.54257813 * 2560; time = 0.0769s; samplesPerSecond = 33306.0
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalErrorPrediction = 0.54648438 * 2560; time = 0.0776s; samplesPerSecond = 32994.4
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalErrorPrediction = 0.53984375 * 2560; time = 0.0765s; samplesPerSecond = 33448.3
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalErrorPrediction = 0.52109375 * 2560; time = 0.0805s; samplesPerSecond = 31791.4
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalErrorPrediction = 0.52148438 * 2560; time = 0.0779s; samplesPerSecond = 32842.8
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalErrorPrediction = 0.51992187 * 2560; time = 0.0797s; samplesPerSecond = 32134.2
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalErrorPrediction = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.670833s
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.08889863 * 2560; EvalClassificationError = 0.56367188 * 2560; time = 0.1034s; samplesPerSecond = 24758.7
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.00776218 * 2560; EvalClassificationError = 0.54218750 * 2560; time = 0.0861s; samplesPerSecond = 29745.7
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 1.99260186 * 2560; EvalClassificationError = 0.54257813 * 2560; time = 0.0769s; samplesPerSecond = 33306.0
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 1.98459919 * 2560; EvalClassificationError = 0.54648438 * 2560; time = 0.0776s; samplesPerSecond = 32994.4
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 1.97206446 * 2560; EvalClassificationError = 0.53984375 * 2560; time = 0.0765s; samplesPerSecond = 33448.3
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 1.91865543 * 2560; EvalClassificationError = 0.52109375 * 2560; time = 0.0805s; samplesPerSecond = 31791.4
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 1.91066650 * 2560; EvalClassificationError = 0.52148438 * 2560; time = 0.0779s; samplesPerSecond = 32842.8
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 2 of 3]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 1.89501438 * 2560; EvalClassificationError = 0.51992187 * 2560; time = 0.0797s; samplesPerSecond = 32134.2
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Finished Epoch[ 2 of 3]: [Training] CrossEntropyWithSoftmax = 1.97128283 * 20480; EvalClassificationError = 0.53715820 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=0.670833s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Starting minibatch loop, DataParallelSGD training (MyRank = 2, NumNodes = 3, NumGradientBits = 32), distributed reading is ENABLED.
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalErrorPrediction = 0.52470703 * 10240; time = 0.1506s; samplesPerSecond = 67980.7
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalErrorPrediction = 0.53974609 * 10240; time = 0.1252s; samplesPerSecond = 81781.3
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalErrorPrediction = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.286844s
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.89820595 * 10240; EvalClassificationError = 0.52470703 * 10240; time = 0.1506s; samplesPerSecond = 67980.7
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Epoch[ 3 of 3]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91958075 * 10240; EvalClassificationError = 0.53974609 * 10240; time = 0.1252s; samplesPerSecond = 81781.3
|
||||
MPI Rank 2: 08/16/2016 03:03:20: Finished Epoch[ 3 of 3]: [Training] CrossEntropyWithSoftmax = 1.90889335 * 20480; EvalClassificationError = 0.53222656 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=0.286844s
|
||||
MPI Rank 2: 08/16/2016 03:03:21: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:21: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -133,7 +133,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -159,7 +159,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -242,7 +242,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -268,7 +268,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -357,7 +357,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -383,7 +383,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -479,7 +479,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -508,7 +508,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -531,7 +531,7 @@ MPI Rank 0: 08/16/2016 09:58:35: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 09:58:35: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:58:35: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 09:58:35: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 09:58:35: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -584,39 +584,39 @@ MPI Rank 0: 08/16/2016 09:58:38: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:58:38: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.2701s; samplesPerSecond = 2369.7
|
||||
MPI Rank 0: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.2438s; samplesPerSecond = 2624.7
|
||||
MPI Rank 0: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.2235s; samplesPerSecond = 2863.0
|
||||
MPI Rank 0: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.2032s; samplesPerSecond = 3149.5
|
||||
MPI Rank 0: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.2198s; samplesPerSecond = 2912.3
|
||||
MPI Rank 0: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.4074s; samplesPerSecond = 1570.8
|
||||
MPI Rank 0: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.2070s; samplesPerSecond = 3091.8
|
||||
MPI Rank 0: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.2217s; samplesPerSecond = 2887.1
|
||||
MPI Rank 0: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.1943s; samplesPerSecond = 3294.6
|
||||
MPI Rank 0: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.2622s; samplesPerSecond = 2440.6
|
||||
MPI Rank 0: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.2782s; samplesPerSecond = 2300.4
|
||||
MPI Rank 0: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.1882s; samplesPerSecond = 3400.0
|
||||
MPI Rank 0: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.2371s; samplesPerSecond = 2699.5
|
||||
MPI Rank 0: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.1719s; samplesPerSecond = 3724.2
|
||||
MPI Rank 0: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.1753s; samplesPerSecond = 3650.7
|
||||
MPI Rank 0: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.3949s; samplesPerSecond = 1620.5
|
||||
MPI Rank 0: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.2098s; samplesPerSecond = 3050.3
|
||||
MPI Rank 0: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.2381s; samplesPerSecond = 2688.4
|
||||
MPI Rank 0: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.2310s; samplesPerSecond = 2770.1
|
||||
MPI Rank 0: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.2781s; samplesPerSecond = 2301.5
|
||||
MPI Rank 0: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.1626s; samplesPerSecond = 3936.0
|
||||
MPI Rank 0: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.2309s; samplesPerSecond = 2771.2
|
||||
MPI Rank 0: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2464s; samplesPerSecond = 2596.9
|
||||
MPI Rank 0: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.4371s; samplesPerSecond = 1464.1
|
||||
MPI Rank 0: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.2429s; samplesPerSecond = 2635.0
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.1963s; samplesPerSecond = 3259.8
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.1323s; samplesPerSecond = 4839.2
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0937s; samplesPerSecond = 6828.6
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.1302s; samplesPerSecond = 4916.5
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.1834s; samplesPerSecond = 3490.3
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.2312s; samplesPerSecond = 2768.5
|
||||
MPI Rank 0: 08/16/2016 09:58:46: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.1764s; samplesPerSecond = 3628.8
|
||||
MPI Rank 0: 08/16/2016 09:58:46: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=7.32475s
|
||||
MPI Rank 0: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.2701s; samplesPerSecond = 2369.7
|
||||
MPI Rank 0: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.2438s; samplesPerSecond = 2624.7
|
||||
MPI Rank 0: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.2235s; samplesPerSecond = 2863.0
|
||||
MPI Rank 0: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.2032s; samplesPerSecond = 3149.5
|
||||
MPI Rank 0: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.2198s; samplesPerSecond = 2912.3
|
||||
MPI Rank 0: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.4074s; samplesPerSecond = 1570.8
|
||||
MPI Rank 0: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.2070s; samplesPerSecond = 3091.8
|
||||
MPI Rank 0: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.2217s; samplesPerSecond = 2887.1
|
||||
MPI Rank 0: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.1943s; samplesPerSecond = 3294.6
|
||||
MPI Rank 0: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.2622s; samplesPerSecond = 2440.6
|
||||
MPI Rank 0: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.2782s; samplesPerSecond = 2300.4
|
||||
MPI Rank 0: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.1882s; samplesPerSecond = 3400.0
|
||||
MPI Rank 0: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.2371s; samplesPerSecond = 2699.5
|
||||
MPI Rank 0: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.1719s; samplesPerSecond = 3724.2
|
||||
MPI Rank 0: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.1753s; samplesPerSecond = 3650.7
|
||||
MPI Rank 0: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.3949s; samplesPerSecond = 1620.5
|
||||
MPI Rank 0: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.2098s; samplesPerSecond = 3050.3
|
||||
MPI Rank 0: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.2381s; samplesPerSecond = 2688.4
|
||||
MPI Rank 0: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.2310s; samplesPerSecond = 2770.1
|
||||
MPI Rank 0: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.2781s; samplesPerSecond = 2301.5
|
||||
MPI Rank 0: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.1626s; samplesPerSecond = 3936.0
|
||||
MPI Rank 0: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.2309s; samplesPerSecond = 2771.2
|
||||
MPI Rank 0: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2464s; samplesPerSecond = 2596.9
|
||||
MPI Rank 0: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.4371s; samplesPerSecond = 1464.1
|
||||
MPI Rank 0: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.2429s; samplesPerSecond = 2635.0
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.1963s; samplesPerSecond = 3259.8
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.1323s; samplesPerSecond = 4839.2
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0937s; samplesPerSecond = 6828.6
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.1302s; samplesPerSecond = 4916.5
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.1834s; samplesPerSecond = 3490.3
|
||||
MPI Rank 0: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.2312s; samplesPerSecond = 2768.5
|
||||
MPI Rank 0: 08/16/2016 09:58:46: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.1764s; samplesPerSecond = 3628.8
|
||||
MPI Rank 0: 08/16/2016 09:58:46: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=7.32475s
|
||||
MPI Rank 0: 08/16/2016 09:58:46: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095713.701165/Speech/DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:58:46: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -626,45 +626,45 @@ MPI Rank 0: 08/16/2016 09:58:46: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.008473
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00554
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.020395
|
||||
MPI Rank 0: 08/16/2016 09:58:46: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17194929 * 2304; EvalErrorPrediction = 0.57942708 * 2304; time = 0.4489s; samplesPerSecond = 5133.1
|
||||
MPI Rank 0: 08/16/2016 09:58:46: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17194929 * 2304; EvalClassificationError = 0.57942708 * 2304; time = 0.4489s; samplesPerSecond = 5133.1
|
||||
MPI Rank 0: Async gradient aggregation wait time: 7e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.058934
|
||||
MPI Rank 0: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.005978
|
||||
MPI Rank 0: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.18854615 * 2560; EvalErrorPrediction = 0.58593750 * 2560; time = 0.5334s; samplesPerSecond = 4799.0
|
||||
MPI Rank 0: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.18854615 * 2560; EvalClassificationError = 0.58593750 * 2560; time = 0.5334s; samplesPerSecond = 4799.0
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.006512
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.051855
|
||||
MPI Rank 0: Async gradient aggregation wait time: 7e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.007858
|
||||
MPI Rank 0: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.21632951 * 2560; EvalErrorPrediction = 0.59179688 * 2560; time = 0.4913s; samplesPerSecond = 5210.6
|
||||
MPI Rank 0: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.21632951 * 2560; EvalClassificationError = 0.59179688 * 2560; time = 0.4913s; samplesPerSecond = 5210.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.139455
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.009705
|
||||
MPI Rank 0: Async gradient aggregation wait time: 8e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.003902
|
||||
MPI Rank 0: 08/16/2016 09:58:48: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.20703856 * 2560; EvalErrorPrediction = 0.58867187 * 2560; time = 0.6431s; samplesPerSecond = 3980.5
|
||||
MPI Rank 0: 08/16/2016 09:58:48: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.20703856 * 2560; EvalClassificationError = 0.58867187 * 2560; time = 0.6431s; samplesPerSecond = 3980.5
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.024974
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.017855
|
||||
MPI Rank 0: 08/16/2016 09:58:49: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.15558289 * 2560; EvalErrorPrediction = 0.58515625 * 2560; time = 0.6357s; samplesPerSecond = 4026.9
|
||||
MPI Rank 0: 08/16/2016 09:58:49: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.15558289 * 2560; EvalClassificationError = 0.58515625 * 2560; time = 0.6357s; samplesPerSecond = 4026.9
|
||||
MPI Rank 0: Async gradient aggregation wait time: 8e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.004211
|
||||
MPI Rank 0: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.007874
|
||||
MPI Rank 0: 08/16/2016 09:58:49: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10682051 * 2560; EvalErrorPrediction = 0.57421875 * 2560; time = 0.5222s; samplesPerSecond = 4902.2
|
||||
MPI Rank 0: 08/16/2016 09:58:49: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10682051 * 2560; EvalClassificationError = 0.57421875 * 2560; time = 0.5222s; samplesPerSecond = 4902.2
|
||||
MPI Rank 0: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.003861
|
||||
MPI Rank 0: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.003945
|
||||
MPI Rank 0: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10935371 * 2560; EvalErrorPrediction = 0.56796875 * 2560; time = 0.6713s; samplesPerSecond = 3813.6
|
||||
MPI Rank 0: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10935371 * 2560; EvalClassificationError = 0.56796875 * 2560; time = 0.6713s; samplesPerSecond = 3813.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.06808
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.010842
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.081552
|
||||
MPI Rank 0: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10066920 * 2560; EvalErrorPrediction = 0.59101563 * 2560; time = 0.6037s; samplesPerSecond = 4240.3
|
||||
MPI Rank 0: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10066920 * 2560; EvalClassificationError = 0.59101563 * 2560; time = 0.6037s; samplesPerSecond = 4240.3
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.011051
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.06767
|
||||
MPI Rank 0: 08/16/2016 09:58:50: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15612951 * 20480; EvalErrorPrediction = 0.58251953 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.63507s
|
||||
MPI Rank 0: 08/16/2016 09:58:50: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15612951 * 20480; EvalClassificationError = 0.58251953 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.63507s
|
||||
MPI Rank 0: 08/16/2016 09:58:50: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095713.701165/Speech/DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:58:51: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -675,13 +675,13 @@ MPI Rank 0: Async gradient aggregation wait time: 1.1e-05
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.009207
|
||||
MPI Rank 0: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.036717
|
||||
MPI Rank 0: 08/16/2016 09:58:53: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.12387153 * 9216; EvalErrorPrediction = 0.56315104 * 9216; time = 1.9738s; samplesPerSecond = 4669.1
|
||||
MPI Rank 0: 08/16/2016 09:58:53: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.12387153 * 9216; EvalClassificationError = 0.56315104 * 9216; time = 1.9738s; samplesPerSecond = 4669.1
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.004899
|
||||
MPI Rank 0: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.004876
|
||||
MPI Rank 0: 08/16/2016 09:58:54: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.00080353 * 10240; EvalErrorPrediction = 0.55566406 * 10240; time = 1.7763s; samplesPerSecond = 5764.8
|
||||
MPI Rank 0: 08/16/2016 09:58:54: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.05662068 * 20480; EvalErrorPrediction = 0.55913086 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.85072s
|
||||
MPI Rank 0: 08/16/2016 09:58:54: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.00080353 * 10240; EvalClassificationError = 0.55566406 * 10240; time = 1.7763s; samplesPerSecond = 5764.8
|
||||
MPI Rank 0: 08/16/2016 09:58:54: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.05662068 * 20480; EvalClassificationError = 0.55913086 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.85072s
|
||||
MPI Rank 0: 08/16/2016 09:58:54: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095713.701165/Speech/DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:58:54: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -692,14 +692,14 @@ MPI Rank 0: Async gradient aggregation wait time: 0.0046
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.069203
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.041271
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.190206
|
||||
MPI Rank 0: 08/16/2016 09:58:55: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94319640 * 9216; EvalErrorPrediction = 0.52680122 * 9216; time = 0.8417s; samplesPerSecond = 10949.3
|
||||
MPI Rank 0: 08/16/2016 09:58:55: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94319640 * 9216; EvalClassificationError = 0.52680122 * 9216; time = 0.8417s; samplesPerSecond = 10949.3
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.07961
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.005129
|
||||
MPI Rank 0: 08/16/2016 09:58:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93779699 * 10240; EvalErrorPrediction = 0.52695313 * 10240; time = 0.6278s; samplesPerSecond = 16312.0
|
||||
MPI Rank 0: 08/16/2016 09:58:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93779699 * 10240; EvalClassificationError = 0.52695313 * 10240; time = 0.6278s; samplesPerSecond = 16312.0
|
||||
MPI Rank 0: Async gradient aggregation wait time: 5e-06
|
||||
MPI Rank 0: 08/16/2016 09:58:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94184346 * 20480; EvalErrorPrediction = 0.52729492 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=1.51956s
|
||||
MPI Rank 0: 08/16/2016 09:58:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94184346 * 20480; EvalClassificationError = 0.52729492 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=1.51956s
|
||||
MPI Rank 0: 08/16/2016 09:58:56: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095713.701165/Speech/DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 09:58:56: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -752,7 +752,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -778,7 +778,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -861,7 +861,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -887,7 +887,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -976,7 +976,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1002,7 +1002,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1098,7 +1098,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1127,7 +1127,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1150,7 +1150,7 @@ MPI Rank 1: 08/16/2016 09:58:35: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 09:58:35: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:35: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 09:58:35: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 09:58:35: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1203,39 +1203,39 @@ MPI Rank 1: 08/16/2016 09:58:38: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:38: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 09:58:38: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.0648s; samplesPerSecond = 9884.0
|
||||
MPI Rank 1: 08/16/2016 09:58:38: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.0519s; samplesPerSecond = 12327.6
|
||||
MPI Rank 1: 08/16/2016 09:58:38: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0525s; samplesPerSecond = 12198.8
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0525s; samplesPerSecond = 12192.6
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.0521s; samplesPerSecond = 12278.9
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.0525s; samplesPerSecond = 12184.2
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.0527s; samplesPerSecond = 12152.5
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.0525s; samplesPerSecond = 12187.7
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.0524s; samplesPerSecond = 12224.2
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.0523s; samplesPerSecond = 12246.2
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.0525s; samplesPerSecond = 12185.6
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.0519s; samplesPerSecond = 12341.9
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0519s; samplesPerSecond = 12341.9
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.0517s; samplesPerSecond = 12384.1
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.0522s; samplesPerSecond = 12253.3
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.0519s; samplesPerSecond = 12323.3
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.0524s; samplesPerSecond = 12219.8
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.0524s; samplesPerSecond = 12216.5
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.0556s; samplesPerSecond = 11509.8
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.0521s; samplesPerSecond = 12292.3
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0521s; samplesPerSecond = 12286.7
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.0528s; samplesPerSecond = 12123.5
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0518s; samplesPerSecond = 12361.7
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.0518s; samplesPerSecond = 12365.2
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0517s; samplesPerSecond = 12370.0
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.0538s; samplesPerSecond = 11888.8
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.0532s; samplesPerSecond = 12023.3
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0520s; samplesPerSecond = 12312.0
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.0522s; samplesPerSecond = 12267.4
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.0534s; samplesPerSecond = 11984.8
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.0530s; samplesPerSecond = 12077.5
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.0526s; samplesPerSecond = 12159.2
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.69586s
|
||||
MPI Rank 1: 08/16/2016 09:58:38: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.0648s; samplesPerSecond = 9884.0
|
||||
MPI Rank 1: 08/16/2016 09:58:38: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.0519s; samplesPerSecond = 12327.6
|
||||
MPI Rank 1: 08/16/2016 09:58:38: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0525s; samplesPerSecond = 12198.8
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0525s; samplesPerSecond = 12192.6
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.0521s; samplesPerSecond = 12278.9
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.0525s; samplesPerSecond = 12184.2
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.0527s; samplesPerSecond = 12152.5
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.0525s; samplesPerSecond = 12187.7
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.0524s; samplesPerSecond = 12224.2
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.0523s; samplesPerSecond = 12246.2
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.0525s; samplesPerSecond = 12185.6
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.0519s; samplesPerSecond = 12341.9
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0519s; samplesPerSecond = 12341.9
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.0517s; samplesPerSecond = 12384.1
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.0522s; samplesPerSecond = 12253.3
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.0519s; samplesPerSecond = 12323.3
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.0524s; samplesPerSecond = 12219.8
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.0524s; samplesPerSecond = 12216.5
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.0556s; samplesPerSecond = 11509.8
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.0521s; samplesPerSecond = 12292.3
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0521s; samplesPerSecond = 12286.7
|
||||
MPI Rank 1: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.0528s; samplesPerSecond = 12123.5
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0518s; samplesPerSecond = 12361.7
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.0518s; samplesPerSecond = 12365.2
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0517s; samplesPerSecond = 12370.0
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.0538s; samplesPerSecond = 11888.8
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.0532s; samplesPerSecond = 12023.3
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0520s; samplesPerSecond = 12312.0
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.0522s; samplesPerSecond = 12267.4
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.0534s; samplesPerSecond = 11984.8
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.0530s; samplesPerSecond = 12077.5
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.0526s; samplesPerSecond = 12159.2
|
||||
MPI Rank 1: 08/16/2016 09:58:40: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=1.69586s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:46: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1244,45 +1244,45 @@ MPI Rank 1: 08/16/2016 09:58:46: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.063039
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.022678
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.025025
|
||||
MPI Rank 1: 08/16/2016 09:58:46: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17194929 * 2304; EvalErrorPrediction = 0.57942708 * 2304; time = 0.4298s; samplesPerSecond = 5360.3
|
||||
MPI Rank 1: 08/16/2016 09:58:46: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17194929 * 2304; EvalClassificationError = 0.57942708 * 2304; time = 0.4298s; samplesPerSecond = 5360.3
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.021114
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.094124
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.01911
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.044389
|
||||
MPI Rank 1: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.18854615 * 2560; EvalErrorPrediction = 0.58593750 * 2560; time = 0.5225s; samplesPerSecond = 4899.8
|
||||
MPI Rank 1: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.18854615 * 2560; EvalClassificationError = 0.58593750 * 2560; time = 0.5225s; samplesPerSecond = 4899.8
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.047583
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.046197
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.054096
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.02731
|
||||
MPI Rank 1: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.21632951 * 2560; EvalErrorPrediction = 0.59179688 * 2560; time = 0.4878s; samplesPerSecond = 5248.4
|
||||
MPI Rank 1: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.21632951 * 2560; EvalClassificationError = 0.59179688 * 2560; time = 0.4878s; samplesPerSecond = 5248.4
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.201925
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.035749
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.038705
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.057511
|
||||
MPI Rank 1: 08/16/2016 09:58:48: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.20703856 * 2560; EvalErrorPrediction = 0.58867187 * 2560; time = 0.6524s; samplesPerSecond = 3924.0
|
||||
MPI Rank 1: 08/16/2016 09:58:48: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.20703856 * 2560; EvalClassificationError = 0.58867187 * 2560; time = 0.6524s; samplesPerSecond = 3924.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.042573
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.054464
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.063915
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.061045
|
||||
MPI Rank 1: 08/16/2016 09:58:48: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.15558289 * 2560; EvalErrorPrediction = 0.58515625 * 2560; time = 0.5263s; samplesPerSecond = 4863.8
|
||||
MPI Rank 1: 08/16/2016 09:58:48: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.15558289 * 2560; EvalClassificationError = 0.58515625 * 2560; time = 0.5263s; samplesPerSecond = 4863.8
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.033596
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.241097
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.051078
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.043873
|
||||
MPI Rank 1: 08/16/2016 09:58:49: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10682051 * 2560; EvalErrorPrediction = 0.57421875 * 2560; time = 0.6232s; samplesPerSecond = 4107.8
|
||||
MPI Rank 1: 08/16/2016 09:58:49: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10682051 * 2560; EvalClassificationError = 0.57421875 * 2560; time = 0.6232s; samplesPerSecond = 4107.8
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.033347
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.032507
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.072162
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.057588
|
||||
MPI Rank 1: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10935371 * 2560; EvalErrorPrediction = 0.56796875 * 2560; time = 0.6658s; samplesPerSecond = 3844.8
|
||||
MPI Rank 1: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10935371 * 2560; EvalClassificationError = 0.56796875 * 2560; time = 0.6658s; samplesPerSecond = 3844.8
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.04021
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.081552
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.055454
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.089611
|
||||
MPI Rank 1: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10066920 * 2560; EvalErrorPrediction = 0.59101563 * 2560; time = 0.5752s; samplesPerSecond = 4450.3
|
||||
MPI Rank 1: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10066920 * 2560; EvalClassificationError = 0.59101563 * 2560; time = 0.5752s; samplesPerSecond = 4450.3
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.07782
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.071704
|
||||
MPI Rank 1: 08/16/2016 09:58:50: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15612951 * 20480; EvalErrorPrediction = 0.58251953 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.63501s
|
||||
MPI Rank 1: 08/16/2016 09:58:50: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15612951 * 20480; EvalClassificationError = 0.58251953 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.63501s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:51: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1292,13 +1292,13 @@ MPI Rank 1: Async gradient aggregation wait time: 0.142203
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.157984
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.179014
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.185782
|
||||
MPI Rank 1: 08/16/2016 09:58:52: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.12387153 * 9216; EvalErrorPrediction = 0.56315104 * 9216; time = 1.8870s; samplesPerSecond = 4883.9
|
||||
MPI Rank 1: 08/16/2016 09:58:52: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.12387153 * 9216; EvalClassificationError = 0.56315104 * 9216; time = 1.8870s; samplesPerSecond = 4883.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.127739
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.082547
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.227062
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.19887
|
||||
MPI Rank 1: 08/16/2016 09:58:54: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.00080353 * 10240; EvalErrorPrediction = 0.55566406 * 10240; time = 1.7231s; samplesPerSecond = 5942.6
|
||||
MPI Rank 1: 08/16/2016 09:58:54: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.05662068 * 20480; EvalErrorPrediction = 0.55913086 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.81748s
|
||||
MPI Rank 1: 08/16/2016 09:58:54: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.00080353 * 10240; EvalClassificationError = 0.55566406 * 10240; time = 1.7231s; samplesPerSecond = 5942.6
|
||||
MPI Rank 1: 08/16/2016 09:58:54: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.05662068 * 20480; EvalClassificationError = 0.55913086 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.81748s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:54: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1308,14 +1308,14 @@ MPI Rank 1: Async gradient aggregation wait time: 0.04561
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.077514
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.069416
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.214632
|
||||
MPI Rank 1: 08/16/2016 09:58:55: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94319640 * 9216; EvalErrorPrediction = 0.52680122 * 9216; time = 0.7835s; samplesPerSecond = 11762.2
|
||||
MPI Rank 1: 08/16/2016 09:58:55: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94319640 * 9216; EvalClassificationError = 0.52680122 * 9216; time = 0.7835s; samplesPerSecond = 11762.2
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.02133
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.100393
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.096348
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.063548
|
||||
MPI Rank 1: 08/16/2016 09:58:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93779699 * 10240; EvalErrorPrediction = 0.52695313 * 10240; time = 0.6536s; samplesPerSecond = 15668.1
|
||||
MPI Rank 1: 08/16/2016 09:58:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93779699 * 10240; EvalClassificationError = 0.52695313 * 10240; time = 0.6536s; samplesPerSecond = 15668.1
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.012294
|
||||
MPI Rank 1: 08/16/2016 09:58:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94184346 * 20480; EvalErrorPrediction = 0.52729492 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=1.51913s
|
||||
MPI Rank 1: 08/16/2016 09:58:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94184346 * 20480; EvalClassificationError = 0.52729492 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=1.51913s
|
||||
MPI Rank 1: 08/16/2016 09:58:56: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:58:56: Action "train" complete.
|
||||
|
|
@ -1367,7 +1367,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1393,7 +1393,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1476,7 +1476,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1502,7 +1502,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1591,7 +1591,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1617,7 +1617,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1713,7 +1713,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1742,7 +1742,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1765,7 +1765,7 @@ MPI Rank 2: 08/16/2016 09:58:36: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 09:58:36: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:58:36: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 09:58:36: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 09:58:36: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1818,39 +1818,39 @@ MPI Rank 2: 08/16/2016 09:58:38: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:58:38: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalErrorPrediction = 0.91093750 * 640; time = 0.4366s; samplesPerSecond = 1465.9
|
||||
MPI Rank 2: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalErrorPrediction = 0.92500000 * 640; time = 0.1664s; samplesPerSecond = 3846.1
|
||||
MPI Rank 2: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.1774s; samplesPerSecond = 3608.3
|
||||
MPI Rank 2: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.1870s; samplesPerSecond = 3423.2
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalErrorPrediction = 0.86250000 * 640; time = 0.2486s; samplesPerSecond = 2574.5
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.1744s; samplesPerSecond = 3669.8
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalErrorPrediction = 0.77500000 * 640; time = 0.1916s; samplesPerSecond = 3340.5
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalErrorPrediction = 0.82812500 * 640; time = 0.1603s; samplesPerSecond = 3992.2
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalErrorPrediction = 0.81875000 * 640; time = 0.1637s; samplesPerSecond = 3910.2
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalErrorPrediction = 0.79843750 * 640; time = 0.1652s; samplesPerSecond = 3873.6
|
||||
MPI Rank 2: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalErrorPrediction = 0.83281250 * 640; time = 0.1545s; samplesPerSecond = 4142.5
|
||||
MPI Rank 2: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalErrorPrediction = 0.79531250 * 640; time = 0.3532s; samplesPerSecond = 1811.9
|
||||
MPI Rank 2: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.1683s; samplesPerSecond = 3802.8
|
||||
MPI Rank 2: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalErrorPrediction = 0.75937500 * 640; time = 0.1955s; samplesPerSecond = 3274.2
|
||||
MPI Rank 2: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.1975s; samplesPerSecond = 3240.7
|
||||
MPI Rank 2: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalErrorPrediction = 0.75000000 * 640; time = 0.2607s; samplesPerSecond = 2455.1
|
||||
MPI Rank 2: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalErrorPrediction = 0.70156250 * 640; time = 0.2756s; samplesPerSecond = 2322.4
|
||||
MPI Rank 2: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalErrorPrediction = 0.65156250 * 640; time = 0.1933s; samplesPerSecond = 3310.1
|
||||
MPI Rank 2: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalErrorPrediction = 0.71250000 * 640; time = 0.2364s; samplesPerSecond = 2706.8
|
||||
MPI Rank 2: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalErrorPrediction = 0.67500000 * 640; time = 0.2414s; samplesPerSecond = 2651.4
|
||||
MPI Rank 2: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.4336s; samplesPerSecond = 1476.1
|
||||
MPI Rank 2: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.1922s; samplesPerSecond = 3329.5
|
||||
MPI Rank 2: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2283s; samplesPerSecond = 2803.0
|
||||
MPI Rank 2: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2306s; samplesPerSecond = 2775.1
|
||||
MPI Rank 2: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.2259s; samplesPerSecond = 2833.1
|
||||
MPI Rank 2: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalErrorPrediction = 0.62343750 * 640; time = 0.2410s; samplesPerSecond = 2656.1
|
||||
MPI Rank 2: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalErrorPrediction = 0.57968750 * 640; time = 0.2429s; samplesPerSecond = 2635.2
|
||||
MPI Rank 2: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1775s; samplesPerSecond = 3604.8
|
||||
MPI Rank 2: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalErrorPrediction = 0.59062500 * 640; time = 0.2432s; samplesPerSecond = 2631.8
|
||||
MPI Rank 2: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.4108s; samplesPerSecond = 1557.9
|
||||
MPI Rank 2: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalErrorPrediction = 0.58281250 * 640; time = 0.1553s; samplesPerSecond = 4121.5
|
||||
MPI Rank 2: 08/16/2016 09:58:46: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalErrorPrediction = 0.60468750 * 640; time = 0.2045s; samplesPerSecond = 3129.6
|
||||
MPI Rank 2: 08/16/2016 09:58:46: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalErrorPrediction = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=7.33875s
|
||||
MPI Rank 2: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.56962759 * 640; EvalClassificationError = 0.91093750 * 640; time = 0.4366s; samplesPerSecond = 1465.9
|
||||
MPI Rank 2: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.33203458 * 640; EvalClassificationError = 0.92500000 * 640; time = 0.1664s; samplesPerSecond = 3846.1
|
||||
MPI Rank 2: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97802531 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.1774s; samplesPerSecond = 3608.3
|
||||
MPI Rank 2: 08/16/2016 09:58:39: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.74456931 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.1870s; samplesPerSecond = 3423.2
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.84496599 * 640; EvalClassificationError = 0.86250000 * 640; time = 0.2486s; samplesPerSecond = 2574.5
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.70662762 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.1744s; samplesPerSecond = 3669.8
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40638941 * 640; EvalClassificationError = 0.77500000 * 640; time = 0.1916s; samplesPerSecond = 3340.5
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51838707 * 640; EvalClassificationError = 0.82812500 * 640; time = 0.1603s; samplesPerSecond = 3992.2
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49900161 * 640; EvalClassificationError = 0.81875000 * 640; time = 0.1637s; samplesPerSecond = 3910.2
|
||||
MPI Rank 2: 08/16/2016 09:58:40: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39551909 * 640; EvalClassificationError = 0.79843750 * 640; time = 0.1652s; samplesPerSecond = 3873.6
|
||||
MPI Rank 2: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.50293318 * 640; EvalClassificationError = 0.83281250 * 640; time = 0.1545s; samplesPerSecond = 4142.5
|
||||
MPI Rank 2: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.27255549 * 640; EvalClassificationError = 0.79531250 * 640; time = 0.3532s; samplesPerSecond = 1811.9
|
||||
MPI Rank 2: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.19692805 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.1683s; samplesPerSecond = 3802.8
|
||||
MPI Rank 2: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.06020026 * 640; EvalClassificationError = 0.75937500 * 640; time = 0.1955s; samplesPerSecond = 3274.2
|
||||
MPI Rank 2: 08/16/2016 09:58:41: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95745162 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.1975s; samplesPerSecond = 3240.7
|
||||
MPI Rank 2: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10529802 * 640; EvalClassificationError = 0.75000000 * 640; time = 0.2607s; samplesPerSecond = 2455.1
|
||||
MPI Rank 2: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.81919831 * 640; EvalClassificationError = 0.70156250 * 640; time = 0.2756s; samplesPerSecond = 2322.4
|
||||
MPI Rank 2: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.71884079 * 640; EvalClassificationError = 0.65156250 * 640; time = 0.1933s; samplesPerSecond = 3310.1
|
||||
MPI Rank 2: 08/16/2016 09:58:42: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.81172687 * 640; EvalClassificationError = 0.71250000 * 640; time = 0.2364s; samplesPerSecond = 2706.8
|
||||
MPI Rank 2: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71883154 * 640; EvalClassificationError = 0.67500000 * 640; time = 0.2414s; samplesPerSecond = 2651.4
|
||||
MPI Rank 2: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.58180764 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.4336s; samplesPerSecond = 1476.1
|
||||
MPI Rank 2: 08/16/2016 09:58:43: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.58988172 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.1922s; samplesPerSecond = 3329.5
|
||||
MPI Rank 2: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51621962 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2283s; samplesPerSecond = 2803.0
|
||||
MPI Rank 2: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.54161790 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2306s; samplesPerSecond = 2775.1
|
||||
MPI Rank 2: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.47626842 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.2259s; samplesPerSecond = 2833.1
|
||||
MPI Rank 2: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.38693259 * 640; EvalClassificationError = 0.62343750 * 640; time = 0.2410s; samplesPerSecond = 2656.1
|
||||
MPI Rank 2: 08/16/2016 09:58:44: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.25003729 * 640; EvalClassificationError = 0.57968750 * 640; time = 0.2429s; samplesPerSecond = 2635.2
|
||||
MPI Rank 2: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.50525264 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1775s; samplesPerSecond = 3604.8
|
||||
MPI Rank 2: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.25697158 * 640; EvalClassificationError = 0.59062500 * 640; time = 0.2432s; samplesPerSecond = 2631.8
|
||||
MPI Rank 2: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.24761175 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.4108s; samplesPerSecond = 1557.9
|
||||
MPI Rank 2: 08/16/2016 09:58:45: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.22232242 * 640; EvalClassificationError = 0.58281250 * 640; time = 0.1553s; samplesPerSecond = 4121.5
|
||||
MPI Rank 2: 08/16/2016 09:58:46: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.22216501 * 640; EvalClassificationError = 0.60468750 * 640; time = 0.2045s; samplesPerSecond = 3129.6
|
||||
MPI Rank 2: 08/16/2016 09:58:46: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.02988126 * 20480; EvalClassificationError = 0.73022461 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=7.33875s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:58:46: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1859,45 +1859,45 @@ MPI Rank 2: 08/16/2016 09:58:46: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.041047
|
||||
MPI Rank 2: Async gradient aggregation wait time: 8e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.004986
|
||||
MPI Rank 2: 08/16/2016 09:58:46: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17194929 * 2304; EvalErrorPrediction = 0.57942708 * 2304; time = 0.4274s; samplesPerSecond = 5390.1
|
||||
MPI Rank 2: 08/16/2016 09:58:46: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.17194929 * 2304; EvalClassificationError = 0.57942708 * 2304; time = 0.4274s; samplesPerSecond = 5390.1
|
||||
MPI Rank 2: Async gradient aggregation wait time: 8e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.076931
|
||||
MPI Rank 2: Async gradient aggregation wait time: 7e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.036044
|
||||
MPI Rank 2: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.18854615 * 2560; EvalErrorPrediction = 0.58593750 * 2560; time = 0.5418s; samplesPerSecond = 4725.3
|
||||
MPI Rank 2: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.18854615 * 2560; EvalClassificationError = 0.58593750 * 2560; time = 0.5418s; samplesPerSecond = 4725.3
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.044057
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.037897
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.02005
|
||||
MPI Rank 2: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.21632951 * 2560; EvalErrorPrediction = 0.59179688 * 2560; time = 0.6676s; samplesPerSecond = 3834.5
|
||||
MPI Rank 2: 08/16/2016 09:58:47: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.21632951 * 2560; EvalClassificationError = 0.59179688 * 2560; time = 0.6676s; samplesPerSecond = 3834.5
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1.2e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.020351
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.044236
|
||||
MPI Rank 2: 08/16/2016 09:58:48: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.20703856 * 2560; EvalErrorPrediction = 0.58867187 * 2560; time = 0.4588s; samplesPerSecond = 5579.7
|
||||
MPI Rank 2: 08/16/2016 09:58:48: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.20703856 * 2560; EvalClassificationError = 0.58867187 * 2560; time = 0.4588s; samplesPerSecond = 5579.7
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.027471
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.050455
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.050343
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.023037
|
||||
MPI Rank 2: 08/16/2016 09:58:48: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.15558289 * 2560; EvalErrorPrediction = 0.58515625 * 2560; time = 0.5342s; samplesPerSecond = 4792.2
|
||||
MPI Rank 2: 08/16/2016 09:58:48: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.15558289 * 2560; EvalClassificationError = 0.58515625 * 2560; time = 0.5342s; samplesPerSecond = 4792.2
|
||||
MPI Rank 2: Async gradient aggregation wait time: 8e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.235132
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.023126
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.039557
|
||||
MPI Rank 2: 08/16/2016 09:58:49: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10682051 * 2560; EvalErrorPrediction = 0.57421875 * 2560; time = 0.6145s; samplesPerSecond = 4166.1
|
||||
MPI Rank 2: 08/16/2016 09:58:49: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10682051 * 2560; EvalClassificationError = 0.57421875 * 2560; time = 0.6145s; samplesPerSecond = 4166.1
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.015202
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.028494
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.036181
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.053585
|
||||
MPI Rank 2: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10935371 * 2560; EvalErrorPrediction = 0.56796875 * 2560; time = 0.6813s; samplesPerSecond = 3757.3
|
||||
MPI Rank 2: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10935371 * 2560; EvalClassificationError = 0.56796875 * 2560; time = 0.6813s; samplesPerSecond = 3757.3
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.005749
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.102056
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004811
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.085594
|
||||
MPI Rank 2: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10066920 * 2560; EvalErrorPrediction = 0.59101563 * 2560; time = 0.5639s; samplesPerSecond = 4540.0
|
||||
MPI Rank 2: 08/16/2016 09:58:50: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.10066920 * 2560; EvalClassificationError = 0.59101563 * 2560; time = 0.5639s; samplesPerSecond = 4540.0
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.069221
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.062729
|
||||
MPI Rank 2: 08/16/2016 09:58:50: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15612951 * 20480; EvalErrorPrediction = 0.58251953 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.63272s
|
||||
MPI Rank 2: 08/16/2016 09:58:50: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.15612951 * 20480; EvalClassificationError = 0.58251953 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.63272s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:58:51: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1907,13 +1907,13 @@ MPI Rank 2: Async gradient aggregation wait time: 9e-06
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.149714
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.120817
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.199767
|
||||
MPI Rank 2: 08/16/2016 09:58:52: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.12387153 * 9216; EvalErrorPrediction = 0.56315104 * 9216; time = 1.9108s; samplesPerSecond = 4823.2
|
||||
MPI Rank 2: 08/16/2016 09:58:52: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.12387153 * 9216; EvalClassificationError = 0.56315104 * 9216; time = 1.9108s; samplesPerSecond = 4823.2
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.04254
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.061903
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.152877
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.196949
|
||||
MPI Rank 2: 08/16/2016 09:58:54: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.00080353 * 10240; EvalErrorPrediction = 0.55566406 * 10240; time = 1.7051s; samplesPerSecond = 6005.5
|
||||
MPI Rank 2: 08/16/2016 09:58:54: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.05662068 * 20480; EvalErrorPrediction = 0.55913086 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.83811s
|
||||
MPI Rank 2: 08/16/2016 09:58:54: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.00080353 * 10240; EvalClassificationError = 0.55566406 * 10240; time = 1.7051s; samplesPerSecond = 6005.5
|
||||
MPI Rank 2: 08/16/2016 09:58:54: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.05662068 * 20480; EvalClassificationError = 0.55913086 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=3.83811s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:58:54: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1923,14 +1923,14 @@ MPI Rank 2: Async gradient aggregation wait time: 9e-06
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.002705
|
||||
MPI Rank 2: Async gradient aggregation wait time: 9e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.003741
|
||||
MPI Rank 2: 08/16/2016 09:58:55: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94319640 * 9216; EvalErrorPrediction = 0.52680122 * 9216; time = 0.7911s; samplesPerSecond = 11649.4
|
||||
MPI Rank 2: 08/16/2016 09:58:55: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.94319640 * 9216; EvalClassificationError = 0.52680122 * 9216; time = 0.7911s; samplesPerSecond = 11649.4
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1.1e-05
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.018113
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.006122
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.067235
|
||||
MPI Rank 2: 08/16/2016 09:58:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93779699 * 10240; EvalErrorPrediction = 0.52695313 * 10240; time = 0.7006s; samplesPerSecond = 14615.2
|
||||
MPI Rank 2: 08/16/2016 09:58:56: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.93779699 * 10240; EvalClassificationError = 0.52695313 * 10240; time = 0.7006s; samplesPerSecond = 14615.2
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.011944
|
||||
MPI Rank 2: 08/16/2016 09:58:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94184346 * 20480; EvalErrorPrediction = 0.52729492 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=1.51787s
|
||||
MPI Rank 2: 08/16/2016 09:58:56: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.94184346 * 20480; EvalClassificationError = 0.52729492 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=1.51787s
|
||||
MPI Rank 2: 08/16/2016 09:58:56: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:58:56: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -133,7 +133,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -159,7 +159,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -242,7 +242,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -268,7 +268,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -357,7 +357,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -383,7 +383,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -480,7 +480,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -509,7 +509,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -532,7 +532,7 @@ MPI Rank 0: 08/16/2016 09:59:01: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 09:59:01: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:01: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 09:59:01: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 09:59:01: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -585,39 +585,39 @@ MPI Rank 0: 08/16/2016 09:59:05: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:05: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 09:59:05: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1005s; samplesPerSecond = 6367.8
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0984s; samplesPerSecond = 6506.3
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0980s; samplesPerSecond = 6527.5
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0979s; samplesPerSecond = 6535.4
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0979s; samplesPerSecond = 6536.1
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0980s; samplesPerSecond = 6532.6
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6525.3
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.1006s; samplesPerSecond = 6360.6
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0989s; samplesPerSecond = 6468.3
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0989s; samplesPerSecond = 6474.3
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.1017s; samplesPerSecond = 6290.5
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0989s; samplesPerSecond = 6473.1
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0989s; samplesPerSecond = 6474.2
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0988s; samplesPerSecond = 6474.5
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6482.1
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0988s; samplesPerSecond = 6479.2
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0989s; samplesPerSecond = 6474.2
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0961s; samplesPerSecond = 6657.9
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0988s; samplesPerSecond = 6478.0
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.1018s; samplesPerSecond = 6287.5
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0989s; samplesPerSecond = 6469.5
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6469.1
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6470.7
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6471.1
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0988s; samplesPerSecond = 6478.8
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0989s; samplesPerSecond = 6468.8
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0961s; samplesPerSecond = 6661.2
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0989s; samplesPerSecond = 6468.5
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.1017s; samplesPerSecond = 6290.2
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0989s; samplesPerSecond = 6468.5
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0989s; samplesPerSecond = 6472.0
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0935s; samplesPerSecond = 6841.9
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16361s
|
||||
MPI Rank 0: 08/16/2016 09:59:05: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1005s; samplesPerSecond = 6367.8
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0984s; samplesPerSecond = 6506.3
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0980s; samplesPerSecond = 6527.5
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0979s; samplesPerSecond = 6535.4
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0979s; samplesPerSecond = 6536.1
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0980s; samplesPerSecond = 6532.6
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6525.3
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.1006s; samplesPerSecond = 6360.6
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0989s; samplesPerSecond = 6468.3
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0989s; samplesPerSecond = 6474.3
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.1017s; samplesPerSecond = 6290.5
|
||||
MPI Rank 0: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0989s; samplesPerSecond = 6473.1
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0989s; samplesPerSecond = 6474.2
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0988s; samplesPerSecond = 6474.5
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6482.1
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0988s; samplesPerSecond = 6479.2
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0989s; samplesPerSecond = 6474.2
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0961s; samplesPerSecond = 6657.9
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0988s; samplesPerSecond = 6478.0
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.1018s; samplesPerSecond = 6287.5
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0989s; samplesPerSecond = 6469.5
|
||||
MPI Rank 0: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6469.1
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6470.7
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6471.1
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0988s; samplesPerSecond = 6478.8
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0989s; samplesPerSecond = 6468.8
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0961s; samplesPerSecond = 6661.2
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0989s; samplesPerSecond = 6468.5
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.1017s; samplesPerSecond = 6290.2
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0989s; samplesPerSecond = 6468.5
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0989s; samplesPerSecond = 6472.0
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0935s; samplesPerSecond = 6841.9
|
||||
MPI Rank 0: 08/16/2016 09:59:08: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16361s
|
||||
MPI Rank 0: 08/16/2016 09:59:09: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095713.701165/Speech/DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -627,45 +627,45 @@ MPI Rank 0: 08/16/2016 09:59:09: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.01227
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00776
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.011351
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.18586881 * 2304; EvalErrorPrediction = 0.58029514 * 2304; time = 0.1372s; samplesPerSecond = 16788.6
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.18586881 * 2304; EvalClassificationError = 0.58029514 * 2304; time = 0.1372s; samplesPerSecond = 16788.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.005855
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.014498
|
||||
MPI Rank 0: Async gradient aggregation wait time: 3e-06
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.00729
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21453123 * 2560; EvalErrorPrediction = 0.59101563 * 2560; time = 0.1358s; samplesPerSecond = 18853.8
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21453123 * 2560; EvalClassificationError = 0.59101563 * 2560; time = 0.1358s; samplesPerSecond = 18853.8
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002716
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.014512
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002917
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.01459
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.23428937 * 2560; EvalErrorPrediction = 0.59843750 * 2560; time = 0.1441s; samplesPerSecond = 17762.6
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.23428937 * 2560; EvalClassificationError = 0.59843750 * 2560; time = 0.1441s; samplesPerSecond = 17762.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002711
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.014346
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002732
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.014237
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.22238576 * 2560; EvalErrorPrediction = 0.59609375 * 2560; time = 0.1450s; samplesPerSecond = 17656.9
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.22238576 * 2560; EvalClassificationError = 0.59609375 * 2560; time = 0.1450s; samplesPerSecond = 17656.9
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00272
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.014389
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.003177
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.014064
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17945944 * 2560; EvalErrorPrediction = 0.58632812 * 2560; time = 0.1446s; samplesPerSecond = 17707.4
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17945944 * 2560; EvalClassificationError = 0.58632812 * 2560; time = 0.1446s; samplesPerSecond = 17707.4
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002666
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.015196
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002854
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.013143
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.13880131 * 2560; EvalErrorPrediction = 0.58164063 * 2560; time = 0.1361s; samplesPerSecond = 18806.4
|
||||
MPI Rank 0: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.13880131 * 2560; EvalClassificationError = 0.58164063 * 2560; time = 0.1361s; samplesPerSecond = 18806.4
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002966
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.014754
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002816
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.014733
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.12741612 * 2560; EvalErrorPrediction = 0.57031250 * 2560; time = 0.1474s; samplesPerSecond = 17368.1
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.12741612 * 2560; EvalClassificationError = 0.57031250 * 2560; time = 0.1474s; samplesPerSecond = 17368.1
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002731
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.014528
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002533
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.012751
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.09486743 * 2560; EvalErrorPrediction = 0.58242187 * 2560; time = 0.1410s; samplesPerSecond = 18153.6
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.09486743 * 2560; EvalClassificationError = 0.58242187 * 2560; time = 0.1410s; samplesPerSecond = 18153.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.004637
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.005743
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.17271296 * 20480; EvalErrorPrediction = 0.58520508 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.14443s
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.17271296 * 20480; EvalClassificationError = 0.58520508 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.14443s
|
||||
MPI Rank 0: 08/16/2016 09:59:10: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095713.701165/Speech/DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -676,13 +676,13 @@ MPI Rank 0: Async gradient aggregation wait time: 0.000306
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.017813
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.003066
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.028051
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.17281503 * 9216; EvalErrorPrediction = 0.55924479 * 9216; time = 0.2685s; samplesPerSecond = 34319.8
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.17281503 * 9216; EvalClassificationError = 0.55924479 * 9216; time = 0.2685s; samplesPerSecond = 34319.8
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002321
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.025119
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.000673
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.020876
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.02446206 * 10240; EvalErrorPrediction = 0.55722656 * 10240; time = 0.2582s; samplesPerSecond = 39663.9
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09074709 * 20480; EvalErrorPrediction = 0.55820313 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.545782s
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.02446206 * 10240; EvalClassificationError = 0.55722656 * 10240; time = 0.2582s; samplesPerSecond = 39663.9
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09074709 * 20480; EvalClassificationError = 0.55820313 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.545782s
|
||||
MPI Rank 0: 08/16/2016 09:59:10: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095713.701165/Speech/DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 09:59:10: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -693,14 +693,14 @@ MPI Rank 0: Async gradient aggregation wait time: 0.003211
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.026824
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.002719
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.026273
|
||||
MPI Rank 0: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95451978 * 9216; EvalErrorPrediction = 0.52962240 * 9216; time = 0.2694s; samplesPerSecond = 34207.2
|
||||
MPI Rank 0: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95451978 * 9216; EvalClassificationError = 0.52962240 * 9216; time = 0.2694s; samplesPerSecond = 34207.2
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.00066
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.026993
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.000862
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.028023
|
||||
MPI Rank 0: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95218466 * 10240; EvalErrorPrediction = 0.52802734 * 10240; time = 0.2695s; samplesPerSecond = 37993.6
|
||||
MPI Rank 0: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95218466 * 10240; EvalClassificationError = 0.52802734 * 10240; time = 0.2695s; samplesPerSecond = 37993.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.004665
|
||||
MPI Rank 0: 08/16/2016 09:59:11: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.95485031 * 20480; EvalErrorPrediction = 0.52915039 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.552277s
|
||||
MPI Rank 0: 08/16/2016 09:59:11: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.95485031 * 20480; EvalClassificationError = 0.52915039 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.552277s
|
||||
MPI Rank 0: 08/16/2016 09:59:11: SGD: Saving checkpoint model '/tmp/cntk-test-20160816095713.701165/Speech/DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_gpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 09:59:11: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -753,7 +753,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -779,7 +779,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -862,7 +862,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -888,7 +888,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -977,7 +977,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -1003,7 +1003,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1100,7 +1100,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1129,7 +1129,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1152,7 +1152,7 @@ MPI Rank 1: 08/16/2016 09:59:01: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 09:59:01: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:01: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 09:59:01: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 09:59:01: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1205,39 +1205,39 @@ MPI Rank 1: 08/16/2016 09:59:05: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:05: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 09:59:05: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1004s; samplesPerSecond = 6376.5
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0977s; samplesPerSecond = 6553.5
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0966s; samplesPerSecond = 6623.3
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0979s; samplesPerSecond = 6535.7
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0979s; samplesPerSecond = 6536.1
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0980s; samplesPerSecond = 6532.6
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6524.6
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.0980s; samplesPerSecond = 6529.1
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0979s; samplesPerSecond = 6539.9
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0988s; samplesPerSecond = 6474.6
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0989s; samplesPerSecond = 6471.2
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0989s; samplesPerSecond = 6473.4
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0988s; samplesPerSecond = 6475.0
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0989s; samplesPerSecond = 6474.2
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6482.4
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0988s; samplesPerSecond = 6479.2
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0988s; samplesPerSecond = 6475.2
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0990s; samplesPerSecond = 6466.8
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0988s; samplesPerSecond = 6477.9
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0990s; samplesPerSecond = 6467.7
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0989s; samplesPerSecond = 6469.9
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6468.8
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6471.4
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6471.0
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0988s; samplesPerSecond = 6478.9
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0989s; samplesPerSecond = 6469.7
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0989s; samplesPerSecond = 6470.5
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0989s; samplesPerSecond = 6469.4
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0989s; samplesPerSecond = 6469.9
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0989s; samplesPerSecond = 6470.3
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0989s; samplesPerSecond = 6472.4
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0989s; samplesPerSecond = 6468.5
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16037s
|
||||
MPI Rank 1: 08/16/2016 09:59:05: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1004s; samplesPerSecond = 6376.5
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0977s; samplesPerSecond = 6553.5
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0966s; samplesPerSecond = 6623.3
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0979s; samplesPerSecond = 6535.7
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0979s; samplesPerSecond = 6536.1
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0980s; samplesPerSecond = 6532.6
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6524.6
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.0980s; samplesPerSecond = 6529.1
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0979s; samplesPerSecond = 6539.9
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0988s; samplesPerSecond = 6474.6
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0989s; samplesPerSecond = 6471.2
|
||||
MPI Rank 1: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0989s; samplesPerSecond = 6473.4
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0988s; samplesPerSecond = 6475.0
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0989s; samplesPerSecond = 6474.2
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6482.4
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0988s; samplesPerSecond = 6479.2
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0988s; samplesPerSecond = 6475.2
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0990s; samplesPerSecond = 6466.8
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0988s; samplesPerSecond = 6477.9
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0990s; samplesPerSecond = 6467.7
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0989s; samplesPerSecond = 6469.9
|
||||
MPI Rank 1: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6468.8
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6471.4
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6471.0
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0988s; samplesPerSecond = 6478.9
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0989s; samplesPerSecond = 6469.7
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0989s; samplesPerSecond = 6470.5
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0989s; samplesPerSecond = 6469.4
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0989s; samplesPerSecond = 6469.9
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0989s; samplesPerSecond = 6470.3
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0989s; samplesPerSecond = 6472.4
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0989s; samplesPerSecond = 6468.5
|
||||
MPI Rank 1: 08/16/2016 09:59:08: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16037s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1246,45 +1246,45 @@ MPI Rank 1: 08/16/2016 09:59:09: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.0237
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.00528
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.011811
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.18586881 * 2304; EvalErrorPrediction = 0.58029514 * 2304; time = 0.1370s; samplesPerSecond = 16815.4
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.18586881 * 2304; EvalClassificationError = 0.58029514 * 2304; time = 0.1370s; samplesPerSecond = 16815.4
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.005125
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.014556
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.0004
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.017647
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21453123 * 2560; EvalErrorPrediction = 0.59101563 * 2560; time = 0.1359s; samplesPerSecond = 18841.5
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21453123 * 2560; EvalClassificationError = 0.59101563 * 2560; time = 0.1359s; samplesPerSecond = 18841.5
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004023
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.014931
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.00307
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.015228
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.23428937 * 2560; EvalErrorPrediction = 0.59843750 * 2560; time = 0.1446s; samplesPerSecond = 17704.0
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.23428937 * 2560; EvalClassificationError = 0.59843750 * 2560; time = 0.1446s; samplesPerSecond = 17704.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004011
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.014453
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004064
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.01441
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.22238576 * 2560; EvalErrorPrediction = 0.59609375 * 2560; time = 0.1450s; samplesPerSecond = 17651.0
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.22238576 * 2560; EvalClassificationError = 0.59609375 * 2560; time = 0.1450s; samplesPerSecond = 17651.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004029
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.014562
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004499
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.014643
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17945944 * 2560; EvalErrorPrediction = 0.58632812 * 2560; time = 0.1448s; samplesPerSecond = 17674.7
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17945944 * 2560; EvalClassificationError = 0.58632812 * 2560; time = 0.1448s; samplesPerSecond = 17674.7
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.001991
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.015141
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003428
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.012517
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.13880131 * 2560; EvalErrorPrediction = 0.58164063 * 2560; time = 0.1356s; samplesPerSecond = 18877.2
|
||||
MPI Rank 1: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.13880131 * 2560; EvalClassificationError = 0.58164063 * 2560; time = 0.1356s; samplesPerSecond = 18877.2
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003845
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.014831
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003709
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.015295
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.12741612 * 2560; EvalErrorPrediction = 0.57031250 * 2560; time = 0.1477s; samplesPerSecond = 17331.6
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.12741612 * 2560; EvalClassificationError = 0.57031250 * 2560; time = 0.1477s; samplesPerSecond = 17331.6
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.00406
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.014925
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004291
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.013058
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.09486743 * 2560; EvalErrorPrediction = 0.58242187 * 2560; time = 0.1411s; samplesPerSecond = 18147.5
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.09486743 * 2560; EvalClassificationError = 0.58242187 * 2560; time = 0.1411s; samplesPerSecond = 18147.5
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.00475
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.005288
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.17271296 * 20480; EvalErrorPrediction = 0.58520508 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.14383s
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.17271296 * 20480; EvalClassificationError = 0.58520508 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.14383s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1294,13 +1294,13 @@ MPI Rank 1: Async gradient aggregation wait time: 0.013339
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.032552
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.007984
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.028121
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.17281503 * 9216; EvalErrorPrediction = 0.55924479 * 9216; time = 0.2689s; samplesPerSecond = 34267.9
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.17281503 * 9216; EvalClassificationError = 0.55924479 * 9216; time = 0.2689s; samplesPerSecond = 34267.9
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.002419
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.026597
|
||||
MPI Rank 1: Async gradient aggregation wait time: 7.7e-05
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.014144
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.02446206 * 10240; EvalErrorPrediction = 0.55722656 * 10240; time = 0.2647s; samplesPerSecond = 38681.1
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09074709 * 20480; EvalErrorPrediction = 0.55820313 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.545761s
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.02446206 * 10240; EvalClassificationError = 0.55722656 * 10240; time = 0.2647s; samplesPerSecond = 38681.1
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09074709 * 20480; EvalClassificationError = 0.55820313 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.545761s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:10: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1310,14 +1310,14 @@ MPI Rank 1: Async gradient aggregation wait time: 0.002841
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.027316
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.002301
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.026786
|
||||
MPI Rank 1: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95451978 * 9216; EvalErrorPrediction = 0.52962240 * 9216; time = 0.2705s; samplesPerSecond = 34067.8
|
||||
MPI Rank 1: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95451978 * 9216; EvalClassificationError = 0.52962240 * 9216; time = 0.2705s; samplesPerSecond = 34067.8
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.00131
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.026665
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.001062
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.02815
|
||||
MPI Rank 1: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95218466 * 10240; EvalErrorPrediction = 0.52802734 * 10240; time = 0.2693s; samplesPerSecond = 38028.7
|
||||
MPI Rank 1: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95218466 * 10240; EvalClassificationError = 0.52802734 * 10240; time = 0.2693s; samplesPerSecond = 38028.7
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004771
|
||||
MPI Rank 1: 08/16/2016 09:59:11: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.95485031 * 20480; EvalErrorPrediction = 0.52915039 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.552241s
|
||||
MPI Rank 1: 08/16/2016 09:59:11: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.95485031 * 20480; EvalClassificationError = 0.52915039 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.552241s
|
||||
MPI Rank 1: 08/16/2016 09:59:11: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 09:59:11: Action "train" complete.
|
||||
|
|
@ -1369,7 +1369,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1395,7 +1395,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1478,7 +1478,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1504,7 +1504,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1593,7 +1593,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1619,7 +1619,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1716,7 +1716,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1745,7 +1745,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1768,7 +1768,7 @@ MPI Rank 2: 08/16/2016 09:59:02: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 09:59:02: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:59:02: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 09:59:02: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 09:59:02: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1821,39 +1821,39 @@ MPI Rank 2: 08/16/2016 09:59:05: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:59:05: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 09:59:05: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalErrorPrediction = 0.94062500 * 640; time = 0.1006s; samplesPerSecond = 6364.4
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalErrorPrediction = 0.92343750 * 640; time = 0.0977s; samplesPerSecond = 6553.8
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalErrorPrediction = 0.89531250 * 640; time = 0.0966s; samplesPerSecond = 6623.4
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.0979s; samplesPerSecond = 6536.1
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalErrorPrediction = 0.88281250 * 640; time = 0.0979s; samplesPerSecond = 6536.9
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalErrorPrediction = 0.86875000 * 640; time = 0.0980s; samplesPerSecond = 6532.9
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalErrorPrediction = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6524.6
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalErrorPrediction = 0.82031250 * 640; time = 0.0980s; samplesPerSecond = 6529.9
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalErrorPrediction = 0.81718750 * 640; time = 0.0979s; samplesPerSecond = 6539.9
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalErrorPrediction = 0.80468750 * 640; time = 0.0988s; samplesPerSecond = 6474.8
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalErrorPrediction = 0.82500000 * 640; time = 0.0989s; samplesPerSecond = 6471.8
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalErrorPrediction = 0.79218750 * 640; time = 0.0989s; samplesPerSecond = 6473.6
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalErrorPrediction = 0.78906250 * 640; time = 0.0988s; samplesPerSecond = 6475.2
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalErrorPrediction = 0.74687500 * 640; time = 0.0989s; samplesPerSecond = 6474.3
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalErrorPrediction = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6483.0
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalErrorPrediction = 0.74062500 * 640; time = 0.0988s; samplesPerSecond = 6479.4
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalErrorPrediction = 0.70625000 * 640; time = 0.0988s; samplesPerSecond = 6475.3
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.0989s; samplesPerSecond = 6469.0
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalErrorPrediction = 0.71718750 * 640; time = 0.0988s; samplesPerSecond = 6476.4
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalErrorPrediction = 0.67812500 * 640; time = 0.0990s; samplesPerSecond = 6467.8
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.0989s; samplesPerSecond = 6469.9
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6469.2
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6471.6
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalErrorPrediction = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6470.9
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalErrorPrediction = 0.64218750 * 640; time = 0.0988s; samplesPerSecond = 6479.3
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalErrorPrediction = 0.62500000 * 640; time = 0.0989s; samplesPerSecond = 6470.1
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.0989s; samplesPerSecond = 6472.2
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalErrorPrediction = 0.65781250 * 640; time = 0.0989s; samplesPerSecond = 6468.0
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalErrorPrediction = 0.59687500 * 640; time = 0.0989s; samplesPerSecond = 6470.0
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalErrorPrediction = 0.60937500 * 640; time = 0.0989s; samplesPerSecond = 6471.1
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalErrorPrediction = 0.58437500 * 640; time = 0.0989s; samplesPerSecond = 6472.4
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalErrorPrediction = 0.60625000 * 640; time = 0.0989s; samplesPerSecond = 6468.6
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalErrorPrediction = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16034s
|
||||
MPI Rank 2: 08/16/2016 09:59:05: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.12%]: CrossEntropyWithSoftmax = 4.62512789 * 640; EvalClassificationError = 0.94062500 * 640; time = 0.1006s; samplesPerSecond = 6364.4
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.35619366 * 640; EvalClassificationError = 0.92343750 * 640; time = 0.0977s; samplesPerSecond = 6553.8
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97911998 * 640; EvalClassificationError = 0.89531250 * 640; time = 0.0966s; samplesPerSecond = 6623.4
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73643568 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.0979s; samplesPerSecond = 6536.1
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.62%]: CrossEntropyWithSoftmax = 3.83079081 * 640; EvalClassificationError = 0.88281250 * 640; time = 0.0979s; samplesPerSecond = 6536.9
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.71437689 * 640; EvalClassificationError = 0.86875000 * 640; time = 0.0980s; samplesPerSecond = 6532.9
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.42186230 * 640; EvalClassificationError = 0.79062500 * 640; time = 0.0981s; samplesPerSecond = 6524.6
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.53658053 * 640; EvalClassificationError = 0.82031250 * 640; time = 0.0980s; samplesPerSecond = 6529.9
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.12%]: CrossEntropyWithSoftmax = 3.49758017 * 640; EvalClassificationError = 0.81718750 * 640; time = 0.0979s; samplesPerSecond = 6539.9
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39996308 * 640; EvalClassificationError = 0.80468750 * 640; time = 0.0988s; samplesPerSecond = 6474.8
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.49445773 * 640; EvalClassificationError = 0.82500000 * 640; time = 0.0989s; samplesPerSecond = 6471.8
|
||||
MPI Rank 2: 08/16/2016 09:59:06: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.26676999 * 640; EvalClassificationError = 0.79218750 * 640; time = 0.0989s; samplesPerSecond = 6473.6
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.62%]: CrossEntropyWithSoftmax = 3.18870173 * 640; EvalClassificationError = 0.78906250 * 640; time = 0.0988s; samplesPerSecond = 6475.2
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.05687264 * 640; EvalClassificationError = 0.74687500 * 640; time = 0.0989s; samplesPerSecond = 6474.3
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.95594569 * 640; EvalClassificationError = 0.71875000 * 640; time = 0.0987s; samplesPerSecond = 6483.0
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.10219604 * 640; EvalClassificationError = 0.74062500 * 640; time = 0.0988s; samplesPerSecond = 6479.4
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.12%]: CrossEntropyWithSoftmax = 2.80745014 * 640; EvalClassificationError = 0.70625000 * 640; time = 0.0988s; samplesPerSecond = 6475.3
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.72061842 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.0989s; samplesPerSecond = 6469.0
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.80425747 * 640; EvalClassificationError = 0.71718750 * 640; time = 0.0988s; samplesPerSecond = 6476.4
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.71253068 * 640; EvalClassificationError = 0.67812500 * 640; time = 0.0990s; samplesPerSecond = 6467.8
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.62%]: CrossEntropyWithSoftmax = 2.59360399 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.0989s; samplesPerSecond = 6469.9
|
||||
MPI Rank 2: 08/16/2016 09:59:07: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.60386649 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6469.2
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.53706678 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6471.6
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.56177343 * 640; EvalClassificationError = 0.65625000 * 640; time = 0.0989s; samplesPerSecond = 6470.9
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.12%]: CrossEntropyWithSoftmax = 2.50118791 * 640; EvalClassificationError = 0.64218750 * 640; time = 0.0988s; samplesPerSecond = 6479.3
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.40119788 * 640; EvalClassificationError = 0.62500000 * 640; time = 0.0989s; samplesPerSecond = 6470.1
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.27491503 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.0989s; samplesPerSecond = 6472.2
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.51724208 * 640; EvalClassificationError = 0.65781250 * 640; time = 0.0989s; samplesPerSecond = 6468.0
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.62%]: CrossEntropyWithSoftmax = 2.27797542 * 640; EvalClassificationError = 0.59687500 * 640; time = 0.0989s; samplesPerSecond = 6470.0
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.26017740 * 640; EvalClassificationError = 0.60937500 * 640; time = 0.0989s; samplesPerSecond = 6471.1
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.24735342 * 640; EvalClassificationError = 0.58437500 * 640; time = 0.0989s; samplesPerSecond = 6472.4
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.23665382 * 640; EvalClassificationError = 0.60625000 * 640; time = 0.0989s; samplesPerSecond = 6468.6
|
||||
MPI Rank 2: 08/16/2016 09:59:08: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.03815141 * 20480; EvalClassificationError = 0.73432617 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=3.16034s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1862,45 +1862,45 @@ MPI Rank 2: 08/16/2016 09:59:09: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.028506
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004749
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.011317
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.18586881 * 2304; EvalErrorPrediction = 0.58029514 * 2304; time = 0.1474s; samplesPerSecond = 15635.1
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.18586881 * 2304; EvalClassificationError = 0.58029514 * 2304; time = 0.1474s; samplesPerSecond = 15635.1
|
||||
MPI Rank 2: Async gradient aggregation wait time: 4e-06
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.007014
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.000736
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.015607
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21453123 * 2560; EvalErrorPrediction = 0.59101563 * 2560; time = 0.1261s; samplesPerSecond = 20303.9
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.21453123 * 2560; EvalClassificationError = 0.59101563 * 2560; time = 0.1261s; samplesPerSecond = 20303.9
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.003993
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.014917
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.003077
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.015162
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.23428937 * 2560; EvalErrorPrediction = 0.59843750 * 2560; time = 0.1442s; samplesPerSecond = 17750.5
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.23428937 * 2560; EvalClassificationError = 0.59843750 * 2560; time = 0.1442s; samplesPerSecond = 17750.5
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.003974
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.014147
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004007
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.014106
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.22238576 * 2560; EvalErrorPrediction = 0.59609375 * 2560; time = 0.1450s; samplesPerSecond = 17655.8
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.22238576 * 2560; EvalClassificationError = 0.59609375 * 2560; time = 0.1450s; samplesPerSecond = 17655.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004019
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.01423
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004455
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.014611
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17945944 * 2560; EvalErrorPrediction = 0.58632812 * 2560; time = 0.1441s; samplesPerSecond = 17763.6
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.17945944 * 2560; EvalClassificationError = 0.58632812 * 2560; time = 0.1441s; samplesPerSecond = 17763.6
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.00371
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.015154
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.000311
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.01123
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.13880131 * 2560; EvalErrorPrediction = 0.58164063 * 2560; time = 0.1365s; samplesPerSecond = 18750.5
|
||||
MPI Rank 2: 08/16/2016 09:59:09: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.13880131 * 2560; EvalClassificationError = 0.58164063 * 2560; time = 0.1365s; samplesPerSecond = 18750.5
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.00417
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.014726
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004062
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.014814
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.12741612 * 2560; EvalErrorPrediction = 0.57031250 * 2560; time = 0.1475s; samplesPerSecond = 17354.5
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.12741612 * 2560; EvalClassificationError = 0.57031250 * 2560; time = 0.1475s; samplesPerSecond = 17354.5
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004031
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.014908
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.002043
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.013253
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.09486743 * 2560; EvalErrorPrediction = 0.58242187 * 2560; time = 0.1410s; samplesPerSecond = 18155.8
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.09486743 * 2560; EvalClassificationError = 0.58242187 * 2560; time = 0.1410s; samplesPerSecond = 18155.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004705
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.005677
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.17271296 * 20480; EvalErrorPrediction = 0.58520508 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.14453s
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.17271296 * 20480; EvalClassificationError = 0.58520508 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=1.14453s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1910,13 +1910,13 @@ MPI Rank 2: Async gradient aggregation wait time: 0.003625
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.032563
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.012159
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.028379
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.17281503 * 9216; EvalErrorPrediction = 0.55924479 * 9216; time = 0.2698s; samplesPerSecond = 34155.1
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.17281503 * 9216; EvalClassificationError = 0.55924479 * 9216; time = 0.2698s; samplesPerSecond = 34155.1
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.006289
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.024908
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004171
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.019629
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.02446206 * 10240; EvalErrorPrediction = 0.55722656 * 10240; time = 0.2584s; samplesPerSecond = 39635.4
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09074709 * 20480; EvalErrorPrediction = 0.55820313 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.546045s
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 2.02446206 * 10240; EvalClassificationError = 0.55722656 * 10240; time = 0.2584s; samplesPerSecond = 39635.4
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.09074709 * 20480; EvalClassificationError = 0.55820313 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-05; epochTime=0.546045s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:59:10: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1926,14 +1926,14 @@ MPI Rank 2: Async gradient aggregation wait time: 0.016234
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.026881
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.005475
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.026302
|
||||
MPI Rank 2: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95451978 * 9216; EvalErrorPrediction = 0.52962240 * 9216; time = 0.2702s; samplesPerSecond = 34110.3
|
||||
MPI Rank 2: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.95451978 * 9216; EvalClassificationError = 0.52962240 * 9216; time = 0.2702s; samplesPerSecond = 34110.3
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004618
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.026388
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004954
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.027853
|
||||
MPI Rank 2: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95218466 * 10240; EvalErrorPrediction = 0.52802734 * 10240; time = 0.2689s; samplesPerSecond = 38086.7
|
||||
MPI Rank 2: 08/16/2016 09:59:11: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.95218466 * 10240; EvalClassificationError = 0.52802734 * 10240; time = 0.2689s; samplesPerSecond = 38086.7
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.004703
|
||||
MPI Rank 2: 08/16/2016 09:59:11: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.95485031 * 20480; EvalErrorPrediction = 0.52915039 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.552485s
|
||||
MPI Rank 2: 08/16/2016 09:59:11: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.95485031 * 20480; EvalClassificationError = 0.52915039 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-05; epochTime=0.552485s
|
||||
MPI Rank 2: 08/16/2016 09:59:11: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 09:59:11: Action "train" complete.
|
||||
|
|
|
|||
|
|
@ -128,7 +128,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -154,7 +154,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -237,7 +237,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -263,7 +263,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -352,7 +352,7 @@ MPI Rank 0: traceLevel = 1
|
|||
MPI Rank 0: SimpleNetworkBuilder = [
|
||||
MPI Rank 0: layerSizes = 363:512:512:132
|
||||
MPI Rank 0: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 0: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 0: evalCriterion = "ClassificationError"
|
||||
MPI Rank 0: layerTypes = "Sigmoid"
|
||||
MPI Rank 0: initValueScale = 1.0
|
||||
MPI Rank 0: applyMeanVarNorm = true
|
||||
|
|
@ -378,7 +378,7 @@ MPI Rank 0: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 0: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 0: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 0: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 0: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 0: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 0: logPrior = LogPrior(labels)
|
||||
MPI Rank 0: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -474,7 +474,7 @@ MPI Rank 0: Post-processing network...
|
|||
MPI Rank 0:
|
||||
MPI Rank 0: 7 roots:
|
||||
MPI Rank 0: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 0: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 0: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 0: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 0: MeanOfFeatures = Mean()
|
||||
MPI Rank 0: PosteriorProb = Softmax()
|
||||
|
|
@ -503,7 +503,7 @@ MPI Rank 0: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 0: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 0: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 0: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 0: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 0: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -526,7 +526,7 @@ MPI Rank 0: 08/16/2016 03:03:40: Training criterion node(s):
|
|||
MPI Rank 0: 08/16/2016 03:03:40: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:40: Evaluation criterion node(s):
|
||||
MPI Rank 0: 08/16/2016 03:03:40: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 0: 08/16/2016 03:03:40: EvalClassificationError = ClassificationError
|
||||
MPI Rank 0:
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -579,39 +579,39 @@ MPI Rank 0: 08/16/2016 03:03:44: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 0: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:44: Starting minibatch loop.
|
||||
MPI Rank 0: 08/16/2016 03:03:44: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.5109s; samplesPerSecond = 1252.7
|
||||
MPI Rank 0: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.5214s; samplesPerSecond = 1227.5
|
||||
MPI Rank 0: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.5097s; samplesPerSecond = 1255.8
|
||||
MPI Rank 0: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.5229s; samplesPerSecond = 1223.9
|
||||
MPI Rank 0: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.5135s; samplesPerSecond = 1246.3
|
||||
MPI Rank 0: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.5062s; samplesPerSecond = 1264.3
|
||||
MPI Rank 0: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.5207s; samplesPerSecond = 1229.2
|
||||
MPI Rank 0: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.5229s; samplesPerSecond = 1224.0
|
||||
MPI Rank 0: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.5191s; samplesPerSecond = 1233.0
|
||||
MPI Rank 0: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.5184s; samplesPerSecond = 1234.6
|
||||
MPI Rank 0: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.5232s; samplesPerSecond = 1223.4
|
||||
MPI Rank 0: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.5020s; samplesPerSecond = 1274.9
|
||||
MPI Rank 0: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.4858s; samplesPerSecond = 1317.5
|
||||
MPI Rank 0: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.2162s; samplesPerSecond = 2960.3
|
||||
MPI Rank 0: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.2164s; samplesPerSecond = 2957.3
|
||||
MPI Rank 0: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.2159s; samplesPerSecond = 2964.9
|
||||
MPI Rank 0: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.2176s; samplesPerSecond = 2941.3
|
||||
MPI Rank 0: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.2186s; samplesPerSecond = 2927.2
|
||||
MPI Rank 0: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.2166s; samplesPerSecond = 2955.0
|
||||
MPI Rank 0: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.2162s; samplesPerSecond = 2960.1
|
||||
MPI Rank 0: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.2164s; samplesPerSecond = 2957.9
|
||||
MPI Rank 0: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.2164s; samplesPerSecond = 2957.1
|
||||
MPI Rank 0: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2163s; samplesPerSecond = 2958.8
|
||||
MPI Rank 0: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.2162s; samplesPerSecond = 2960.0
|
||||
MPI Rank 0: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.2170s; samplesPerSecond = 2949.4
|
||||
MPI Rank 0: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.2182s; samplesPerSecond = 2932.7
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.1228s; samplesPerSecond = 5211.7
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.1195s; samplesPerSecond = 5354.4
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.1191s; samplesPerSecond = 5372.5
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.1218s; samplesPerSecond = 5253.5
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.1217s; samplesPerSecond = 5258.4
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.1194s; samplesPerSecond = 5362.1
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=10.2322s
|
||||
MPI Rank 0: 08/16/2016 03:03:44: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.5109s; samplesPerSecond = 1252.7
|
||||
MPI Rank 0: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.5214s; samplesPerSecond = 1227.5
|
||||
MPI Rank 0: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.5097s; samplesPerSecond = 1255.8
|
||||
MPI Rank 0: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.5229s; samplesPerSecond = 1223.9
|
||||
MPI Rank 0: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.5135s; samplesPerSecond = 1246.3
|
||||
MPI Rank 0: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.5062s; samplesPerSecond = 1264.3
|
||||
MPI Rank 0: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.5207s; samplesPerSecond = 1229.2
|
||||
MPI Rank 0: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.5229s; samplesPerSecond = 1224.0
|
||||
MPI Rank 0: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.5191s; samplesPerSecond = 1233.0
|
||||
MPI Rank 0: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.5184s; samplesPerSecond = 1234.6
|
||||
MPI Rank 0: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.5232s; samplesPerSecond = 1223.4
|
||||
MPI Rank 0: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.5020s; samplesPerSecond = 1274.9
|
||||
MPI Rank 0: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.4858s; samplesPerSecond = 1317.5
|
||||
MPI Rank 0: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.2162s; samplesPerSecond = 2960.3
|
||||
MPI Rank 0: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.2164s; samplesPerSecond = 2957.3
|
||||
MPI Rank 0: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.2159s; samplesPerSecond = 2964.9
|
||||
MPI Rank 0: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.2176s; samplesPerSecond = 2941.3
|
||||
MPI Rank 0: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.2186s; samplesPerSecond = 2927.2
|
||||
MPI Rank 0: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.2166s; samplesPerSecond = 2955.0
|
||||
MPI Rank 0: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.2162s; samplesPerSecond = 2960.1
|
||||
MPI Rank 0: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.2164s; samplesPerSecond = 2957.9
|
||||
MPI Rank 0: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.2164s; samplesPerSecond = 2957.1
|
||||
MPI Rank 0: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2163s; samplesPerSecond = 2958.8
|
||||
MPI Rank 0: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.2162s; samplesPerSecond = 2960.0
|
||||
MPI Rank 0: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.2170s; samplesPerSecond = 2949.4
|
||||
MPI Rank 0: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.2182s; samplesPerSecond = 2932.7
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.1228s; samplesPerSecond = 5211.7
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.1195s; samplesPerSecond = 5354.4
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.1191s; samplesPerSecond = 5372.5
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.1218s; samplesPerSecond = 5253.5
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.1217s; samplesPerSecond = 5258.4
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.1194s; samplesPerSecond = 5362.1
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=10.2322s
|
||||
MPI Rank 0: 08/16/2016 03:03:54: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030157.855216\Speech\DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.1'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:54: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
|
|
@ -621,45 +621,45 @@ MPI Rank 0: 08/16/2016 03:03:54: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.010875
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.046041
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.070247
|
||||
MPI Rank 0: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.14273439 * 2304; EvalErrorPrediction = 0.57074653 * 2304; time = 0.5755s; samplesPerSecond = 4003.6
|
||||
MPI Rank 0: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.14273439 * 2304; EvalClassificationError = 0.57074653 * 2304; time = 0.5755s; samplesPerSecond = 4003.6
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.008658
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.080788
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.007539
|
||||
MPI Rank 0: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.16748503 * 2560; EvalErrorPrediction = 0.58085937 * 2560; time = 0.6397s; samplesPerSecond = 4001.9
|
||||
MPI Rank 0: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.16748503 * 2560; EvalClassificationError = 0.58085937 * 2560; time = 0.6397s; samplesPerSecond = 4001.9
|
||||
MPI Rank 0: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.007879
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.023341
|
||||
MPI Rank 0: 08/16/2016 03:03:56: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.20296428 * 2560; EvalErrorPrediction = 0.58945313 * 2560; time = 0.5117s; samplesPerSecond = 5003.1
|
||||
MPI Rank 0: 08/16/2016 03:03:56: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.20296428 * 2560; EvalClassificationError = 0.58945313 * 2560; time = 0.5117s; samplesPerSecond = 5003.1
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.032881
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.007624
|
||||
MPI Rank 0: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.19897771 * 2560; EvalErrorPrediction = 0.58867187 * 2560; time = 0.5514s; samplesPerSecond = 4642.4
|
||||
MPI Rank 0: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.19897771 * 2560; EvalClassificationError = 0.58867187 * 2560; time = 0.5514s; samplesPerSecond = 4642.4
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.052196
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.007301
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.061666
|
||||
MPI Rank 0: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.13999099 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.5867s; samplesPerSecond = 4363.1
|
||||
MPI Rank 0: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.13999099 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.5867s; samplesPerSecond = 4363.1
|
||||
MPI Rank 0: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.054634
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.008597
|
||||
MPI Rank 0: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10194078 * 2560; EvalErrorPrediction = 0.56875000 * 2560; time = 0.5781s; samplesPerSecond = 4428.4
|
||||
MPI Rank 0: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10194078 * 2560; EvalClassificationError = 0.56875000 * 2560; time = 0.5781s; samplesPerSecond = 4428.4
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.008929
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.071019
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.011728
|
||||
MPI Rank 0: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10371395 * 2560; EvalErrorPrediction = 0.56328125 * 2560; time = 0.5530s; samplesPerSecond = 4629.2
|
||||
MPI Rank 0: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10371395 * 2560; EvalClassificationError = 0.56328125 * 2560; time = 0.5530s; samplesPerSecond = 4629.2
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.008518
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.063783
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.064043
|
||||
MPI Rank 0: 08/16/2016 03:03:59: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.08426773 * 2560; EvalErrorPrediction = 0.58398438 * 2560; time = 0.5975s; samplesPerSecond = 4284.3
|
||||
MPI Rank 0: 08/16/2016 03:03:59: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.08426773 * 2560; EvalClassificationError = 0.58398438 * 2560; time = 0.5975s; samplesPerSecond = 4284.3
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.006574
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.03026
|
||||
MPI Rank 0: 08/16/2016 03:03:59: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14176518 * 20480; EvalErrorPrediction = 0.57827148 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.64158s
|
||||
MPI Rank 0: 08/16/2016 03:03:59: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14176518 * 20480; EvalClassificationError = 0.57827148 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.64158s
|
||||
MPI Rank 0: 08/16/2016 03:03:59: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030157.855216\Speech\DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.2'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:03:59: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -670,13 +670,13 @@ MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.106492
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.115088
|
||||
MPI Rank 0: 08/16/2016 03:04:00: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11704698 * 9216; EvalErrorPrediction = 0.55631510 * 9216; time = 1.4653s; samplesPerSecond = 6289.4
|
||||
MPI Rank 0: 08/16/2016 03:04:00: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11704698 * 9216; EvalClassificationError = 0.55631510 * 9216; time = 1.4653s; samplesPerSecond = 6289.4
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.025592
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.164706
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.106746
|
||||
MPI Rank 0: 08/16/2016 03:04:02: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97705226 * 10240; EvalErrorPrediction = 0.54746094 * 10240; time = 1.4775s; samplesPerSecond = 6930.5
|
||||
MPI Rank 0: 08/16/2016 03:04:02: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.04081650 * 20480; EvalErrorPrediction = 0.55122070 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=3.06919s
|
||||
MPI Rank 0: 08/16/2016 03:04:02: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97705226 * 10240; EvalClassificationError = 0.54746094 * 10240; time = 1.4775s; samplesPerSecond = 6930.5
|
||||
MPI Rank 0: 08/16/2016 03:04:02: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.04081650 * 20480; EvalClassificationError = 0.55122070 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=3.06919s
|
||||
MPI Rank 0: 08/16/2016 03:04:02: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030157.855216\Speech\DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn.3'
|
||||
MPI Rank 0:
|
||||
MPI Rank 0: 08/16/2016 03:04:02: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
|
|
@ -687,14 +687,14 @@ MPI Rank 0: Async gradient aggregation wait time: 3e-006
|
|||
MPI Rank 0: Actual gradient aggregation time: 0.010023
|
||||
MPI Rank 0: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.017687
|
||||
MPI Rank 0: 08/16/2016 03:04:04: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.92306138 * 9216; EvalErrorPrediction = 0.52398003 * 9216; time = 1.3966s; samplesPerSecond = 6598.9
|
||||
MPI Rank 0: 08/16/2016 03:04:04: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.92306138 * 9216; EvalClassificationError = 0.52398003 * 9216; time = 1.3966s; samplesPerSecond = 6598.9
|
||||
MPI Rank 0: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.032761
|
||||
MPI Rank 0: Async gradient aggregation wait time: 0.023301
|
||||
MPI Rank 0: Actual gradient aggregation time: 0.055983
|
||||
MPI Rank 0: 08/16/2016 03:04:05: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91800591 * 10240; EvalErrorPrediction = 0.52324219 * 10240; time = 1.5814s; samplesPerSecond = 6475.4
|
||||
MPI Rank 0: 08/16/2016 03:04:05: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91800591 * 10240; EvalClassificationError = 0.52324219 * 10240; time = 1.5814s; samplesPerSecond = 6475.4
|
||||
MPI Rank 0: Async gradient aggregation wait time: 1e-006
|
||||
MPI Rank 0: 08/16/2016 03:04:05: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.92214429 * 20480; EvalErrorPrediction = 0.52407227 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=3.10398s
|
||||
MPI Rank 0: 08/16/2016 03:04:05: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.92214429 * 20480; EvalClassificationError = 0.52407227 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=3.10398s
|
||||
MPI Rank 0: 08/16/2016 03:04:05: SGD: Saving checkpoint model 'C:\Users\svcphil\AppData\Local\Temp\cntk-test-20160816030157.855216\Speech\DNN_ParallelNoQuantizationBufferedAsyncGradientAggregation@release_cpu/models/cntkSpeech.dnn'
|
||||
MPI Rank 0: 08/16/2016 03:04:05: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 0:
|
||||
|
|
@ -745,7 +745,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -771,7 +771,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -854,7 +854,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -880,7 +880,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -969,7 +969,7 @@ MPI Rank 1: traceLevel = 1
|
|||
MPI Rank 1: SimpleNetworkBuilder = [
|
||||
MPI Rank 1: layerSizes = 363:512:512:132
|
||||
MPI Rank 1: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 1: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 1: evalCriterion = "ClassificationError"
|
||||
MPI Rank 1: layerTypes = "Sigmoid"
|
||||
MPI Rank 1: initValueScale = 1.0
|
||||
MPI Rank 1: applyMeanVarNorm = true
|
||||
|
|
@ -995,7 +995,7 @@ MPI Rank 1: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 1: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 1: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 1: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 1: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 1: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 1: logPrior = LogPrior(labels)
|
||||
MPI Rank 1: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1091,7 +1091,7 @@ MPI Rank 1: Post-processing network...
|
|||
MPI Rank 1:
|
||||
MPI Rank 1: 7 roots:
|
||||
MPI Rank 1: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 1: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 1: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 1: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 1: MeanOfFeatures = Mean()
|
||||
MPI Rank 1: PosteriorProb = Softmax()
|
||||
|
|
@ -1120,7 +1120,7 @@ MPI Rank 1: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 1: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 1: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 1: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 1: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 1: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1143,7 +1143,7 @@ MPI Rank 1: 08/16/2016 03:03:41: Training criterion node(s):
|
|||
MPI Rank 1: 08/16/2016 03:03:41: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:41: Evaluation criterion node(s):
|
||||
MPI Rank 1: 08/16/2016 03:03:41: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 1: 08/16/2016 03:03:41: EvalClassificationError = ClassificationError
|
||||
MPI Rank 1:
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1196,39 +1196,39 @@ MPI Rank 1: 08/16/2016 03:03:44: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 1: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:44: Starting minibatch loop.
|
||||
MPI Rank 1: 08/16/2016 03:03:44: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.3609s; samplesPerSecond = 1773.5
|
||||
MPI Rank 1: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.3416s; samplesPerSecond = 1873.3
|
||||
MPI Rank 1: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.3565s; samplesPerSecond = 1795.1
|
||||
MPI Rank 1: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.3512s; samplesPerSecond = 1822.3
|
||||
MPI Rank 1: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.3550s; samplesPerSecond = 1802.7
|
||||
MPI Rank 1: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.3406s; samplesPerSecond = 1879.0
|
||||
MPI Rank 1: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.3527s; samplesPerSecond = 1814.7
|
||||
MPI Rank 1: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.3538s; samplesPerSecond = 1808.7
|
||||
MPI Rank 1: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.3496s; samplesPerSecond = 1830.6
|
||||
MPI Rank 1: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.3463s; samplesPerSecond = 1848.3
|
||||
MPI Rank 1: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.3474s; samplesPerSecond = 1842.2
|
||||
MPI Rank 1: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.3475s; samplesPerSecond = 1841.9
|
||||
MPI Rank 1: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.3556s; samplesPerSecond = 1799.6
|
||||
MPI Rank 1: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.3557s; samplesPerSecond = 1799.3
|
||||
MPI Rank 1: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.3396s; samplesPerSecond = 1884.8
|
||||
MPI Rank 1: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.3504s; samplesPerSecond = 1826.6
|
||||
MPI Rank 1: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.3566s; samplesPerSecond = 1794.7
|
||||
MPI Rank 1: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.3556s; samplesPerSecond = 1799.9
|
||||
MPI Rank 1: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.3482s; samplesPerSecond = 1838.0
|
||||
MPI Rank 1: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.2173s; samplesPerSecond = 2945.4
|
||||
MPI Rank 1: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.2174s; samplesPerSecond = 2943.6
|
||||
MPI Rank 1: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.2171s; samplesPerSecond = 2948.3
|
||||
MPI Rank 1: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2175s; samplesPerSecond = 2943.1
|
||||
MPI Rank 1: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.2185s; samplesPerSecond = 2929.4
|
||||
MPI Rank 1: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.2165s; samplesPerSecond = 2956.2
|
||||
MPI Rank 1: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.2162s; samplesPerSecond = 2960.3
|
||||
MPI Rank 1: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.2164s; samplesPerSecond = 2957.8
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.2164s; samplesPerSecond = 2957.0
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.2163s; samplesPerSecond = 2958.9
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.2162s; samplesPerSecond = 2959.9
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.2163s; samplesPerSecond = 2959.1
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.2195s; samplesPerSecond = 2916.1
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=9.50159s
|
||||
MPI Rank 1: 08/16/2016 03:03:44: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.3609s; samplesPerSecond = 1773.5
|
||||
MPI Rank 1: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.3416s; samplesPerSecond = 1873.3
|
||||
MPI Rank 1: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.3565s; samplesPerSecond = 1795.1
|
||||
MPI Rank 1: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.3512s; samplesPerSecond = 1822.3
|
||||
MPI Rank 1: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.3550s; samplesPerSecond = 1802.7
|
||||
MPI Rank 1: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.3406s; samplesPerSecond = 1879.0
|
||||
MPI Rank 1: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.3527s; samplesPerSecond = 1814.7
|
||||
MPI Rank 1: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.3538s; samplesPerSecond = 1808.7
|
||||
MPI Rank 1: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.3496s; samplesPerSecond = 1830.6
|
||||
MPI Rank 1: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.3463s; samplesPerSecond = 1848.3
|
||||
MPI Rank 1: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.3474s; samplesPerSecond = 1842.2
|
||||
MPI Rank 1: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.3475s; samplesPerSecond = 1841.9
|
||||
MPI Rank 1: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.3556s; samplesPerSecond = 1799.6
|
||||
MPI Rank 1: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.3557s; samplesPerSecond = 1799.3
|
||||
MPI Rank 1: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.3396s; samplesPerSecond = 1884.8
|
||||
MPI Rank 1: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.3504s; samplesPerSecond = 1826.6
|
||||
MPI Rank 1: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.3566s; samplesPerSecond = 1794.7
|
||||
MPI Rank 1: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.3556s; samplesPerSecond = 1799.9
|
||||
MPI Rank 1: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.3482s; samplesPerSecond = 1838.0
|
||||
MPI Rank 1: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.2173s; samplesPerSecond = 2945.4
|
||||
MPI Rank 1: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.2174s; samplesPerSecond = 2943.6
|
||||
MPI Rank 1: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.2171s; samplesPerSecond = 2948.3
|
||||
MPI Rank 1: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2175s; samplesPerSecond = 2943.1
|
||||
MPI Rank 1: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.2185s; samplesPerSecond = 2929.4
|
||||
MPI Rank 1: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.2165s; samplesPerSecond = 2956.2
|
||||
MPI Rank 1: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.2162s; samplesPerSecond = 2960.3
|
||||
MPI Rank 1: 08/16/2016 03:03:52: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.2164s; samplesPerSecond = 2957.8
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.2164s; samplesPerSecond = 2957.0
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.2163s; samplesPerSecond = 2958.9
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.2162s; samplesPerSecond = 2959.9
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.2163s; samplesPerSecond = 2959.1
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.2195s; samplesPerSecond = 2916.1
|
||||
MPI Rank 1: 08/16/2016 03:03:53: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=9.50159s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:54: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1237,45 +1237,45 @@ MPI Rank 1: 08/16/2016 03:03:54: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.033393
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.005092
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.070288
|
||||
MPI Rank 1: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.14273439 * 2304; EvalErrorPrediction = 0.57074653 * 2304; time = 0.5942s; samplesPerSecond = 3877.6
|
||||
MPI Rank 1: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.14273439 * 2304; EvalClassificationError = 0.57074653 * 2304; time = 0.5942s; samplesPerSecond = 3877.6
|
||||
MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.066943
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.013808
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.044491
|
||||
MPI Rank 1: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.16748503 * 2560; EvalErrorPrediction = 0.58085937 * 2560; time = 0.5630s; samplesPerSecond = 4547.4
|
||||
MPI Rank 1: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.16748503 * 2560; EvalClassificationError = 0.58085937 * 2560; time = 0.5630s; samplesPerSecond = 4547.4
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.010263
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.040831
|
||||
MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.056003
|
||||
MPI Rank 1: 08/16/2016 03:03:56: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.20296428 * 2560; EvalErrorPrediction = 0.58945313 * 2560; time = 0.5873s; samplesPerSecond = 4358.6
|
||||
MPI Rank 1: 08/16/2016 03:03:56: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.20296428 * 2560; EvalClassificationError = 0.58945313 * 2560; time = 0.5873s; samplesPerSecond = 4358.6
|
||||
MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.007842
|
||||
MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.0423
|
||||
MPI Rank 1: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.19897771 * 2560; EvalErrorPrediction = 0.58867187 * 2560; time = 0.5887s; samplesPerSecond = 4348.5
|
||||
MPI Rank 1: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.19897771 * 2560; EvalClassificationError = 0.58867187 * 2560; time = 0.5887s; samplesPerSecond = 4348.5
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.004383
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.007521
|
||||
MPI Rank 1: Async gradient aggregation wait time: 5e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.046436
|
||||
MPI Rank 1: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.13999099 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.5709s; samplesPerSecond = 4484.4
|
||||
MPI Rank 1: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.13999099 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.5709s; samplesPerSecond = 4484.4
|
||||
MPI Rank 1: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.043155
|
||||
MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.034426
|
||||
MPI Rank 1: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10194078 * 2560; EvalErrorPrediction = 0.56875000 * 2560; time = 0.5726s; samplesPerSecond = 4470.8
|
||||
MPI Rank 1: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10194078 * 2560; EvalClassificationError = 0.56875000 * 2560; time = 0.5726s; samplesPerSecond = 4470.8
|
||||
MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.049226
|
||||
MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.008603
|
||||
MPI Rank 1: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10371395 * 2560; EvalErrorPrediction = 0.56328125 * 2560; time = 0.5545s; samplesPerSecond = 4617.0
|
||||
MPI Rank 1: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10371395 * 2560; EvalClassificationError = 0.56328125 * 2560; time = 0.5545s; samplesPerSecond = 4617.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.047434
|
||||
MPI Rank 1: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.048141
|
||||
MPI Rank 1: 08/16/2016 03:03:59: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.08426773 * 2560; EvalErrorPrediction = 0.58398438 * 2560; time = 0.5780s; samplesPerSecond = 4429.0
|
||||
MPI Rank 1: 08/16/2016 03:03:59: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.08426773 * 2560; EvalClassificationError = 0.58398438 * 2560; time = 0.5780s; samplesPerSecond = 4429.0
|
||||
MPI Rank 1: Async gradient aggregation wait time: 1e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.021377
|
||||
MPI Rank 1: 08/16/2016 03:03:59: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14176518 * 20480; EvalErrorPrediction = 0.57827148 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.63879s
|
||||
MPI Rank 1: 08/16/2016 03:03:59: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14176518 * 20480; EvalClassificationError = 0.57827148 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.63879s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:03:59: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1285,13 +1285,13 @@ MPI Rank 1: Async gradient aggregation wait time: 0.056538
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.181181
|
||||
MPI Rank 1: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.180513
|
||||
MPI Rank 1: 08/16/2016 03:04:00: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11704698 * 9216; EvalErrorPrediction = 0.55631510 * 9216; time = 1.4418s; samplesPerSecond = 6392.1
|
||||
MPI Rank 1: 08/16/2016 03:04:00: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11704698 * 9216; EvalClassificationError = 0.55631510 * 9216; time = 1.4418s; samplesPerSecond = 6392.1
|
||||
MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.147884
|
||||
MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.032881
|
||||
MPI Rank 1: 08/16/2016 03:04:02: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97705226 * 10240; EvalErrorPrediction = 0.54746094 * 10240; time = 1.6004s; samplesPerSecond = 6398.5
|
||||
MPI Rank 1: 08/16/2016 03:04:02: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.04081650 * 20480; EvalErrorPrediction = 0.55122070 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=3.0669s
|
||||
MPI Rank 1: 08/16/2016 03:04:02: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97705226 * 10240; EvalClassificationError = 0.54746094 * 10240; time = 1.6004s; samplesPerSecond = 6398.5
|
||||
MPI Rank 1: 08/16/2016 03:04:02: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.04081650 * 20480; EvalClassificationError = 0.55122070 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=3.0669s
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:04:02: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 1: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 1 of 3, with 1 datapasses
|
||||
|
|
@ -1301,14 +1301,14 @@ MPI Rank 1: Async gradient aggregation wait time: 2e-006
|
|||
MPI Rank 1: Actual gradient aggregation time: 0.028628
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.062352
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.08373
|
||||
MPI Rank 1: 08/16/2016 03:04:04: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.92306138 * 9216; EvalErrorPrediction = 0.52398003 * 9216; time = 1.4882s; samplesPerSecond = 6192.7
|
||||
MPI Rank 1: 08/16/2016 03:04:04: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.92306138 * 9216; EvalClassificationError = 0.52398003 * 9216; time = 1.4882s; samplesPerSecond = 6192.7
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003042
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.055664
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.003145
|
||||
MPI Rank 1: Actual gradient aggregation time: 0.056467
|
||||
MPI Rank 1: 08/16/2016 03:04:05: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91800591 * 10240; EvalErrorPrediction = 0.52324219 * 10240; time = 1.5850s; samplesPerSecond = 6460.4
|
||||
MPI Rank 1: 08/16/2016 03:04:05: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91800591 * 10240; EvalClassificationError = 0.52324219 * 10240; time = 1.5850s; samplesPerSecond = 6460.4
|
||||
MPI Rank 1: Async gradient aggregation wait time: 0.00839
|
||||
MPI Rank 1: 08/16/2016 03:04:05: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.92214429 * 20480; EvalErrorPrediction = 0.52407227 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=3.10017s
|
||||
MPI Rank 1: 08/16/2016 03:04:05: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.92214429 * 20480; EvalClassificationError = 0.52407227 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=3.10017s
|
||||
MPI Rank 1: 08/16/2016 03:04:05: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 1:
|
||||
MPI Rank 1: 08/16/2016 03:04:05: Action "train" complete.
|
||||
|
|
@ -1358,7 +1358,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1384,7 +1384,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1467,7 +1467,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1493,7 +1493,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1582,7 +1582,7 @@ MPI Rank 2: traceLevel = 1
|
|||
MPI Rank 2: SimpleNetworkBuilder = [
|
||||
MPI Rank 2: layerSizes = 363:512:512:132
|
||||
MPI Rank 2: trainingCriterion = "CrossEntropyWithSoftmax"
|
||||
MPI Rank 2: evalCriterion = "ErrorPrediction"
|
||||
MPI Rank 2: evalCriterion = "ClassificationError"
|
||||
MPI Rank 2: layerTypes = "Sigmoid"
|
||||
MPI Rank 2: initValueScale = 1.0
|
||||
MPI Rank 2: applyMeanVarNorm = true
|
||||
|
|
@ -1608,7 +1608,7 @@ MPI Rank 2: CE = if trainingCriterion == 'CE'
|
|||
MPI Rank 2: then CrossEntropyWithSoftmax(labels, outZ, tag='criterion')
|
||||
MPI Rank 2: else Fail('unknown trainingCriterion ' + trainingCriterion)
|
||||
MPI Rank 2: Err = if evalCriterion == 'Err' then
|
||||
MPI Rank 2: ErrorPrediction(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: ClassificationError(labels, outZ, tag='evaluation')
|
||||
MPI Rank 2: else Fail('unknown evalCriterion ' + evalCriterion)
|
||||
MPI Rank 2: logPrior = LogPrior(labels)
|
||||
MPI Rank 2: // TODO: how to add a tag to an infix operation?
|
||||
|
|
@ -1704,7 +1704,7 @@ MPI Rank 2: Post-processing network...
|
|||
MPI Rank 2:
|
||||
MPI Rank 2: 7 roots:
|
||||
MPI Rank 2: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax()
|
||||
MPI Rank 2: EvalErrorPrediction = ErrorPrediction()
|
||||
MPI Rank 2: EvalClassificationError = ClassificationError()
|
||||
MPI Rank 2: InvStdOfFeatures = InvStdDev()
|
||||
MPI Rank 2: MeanOfFeatures = Mean()
|
||||
MPI Rank 2: PosteriorProb = Softmax()
|
||||
|
|
@ -1733,7 +1733,7 @@ MPI Rank 2: Validating --> W2*H1 = Times (W2, H2) : [132 x 512], [512 x 1 x *] -
|
|||
MPI Rank 2: Validating --> B2 = LearnableParameter() : -> [132 x 1]
|
||||
MPI Rank 2: Validating --> HLast = Plus (W2*H1, B2) : [132 x 1 x *], [132 x 1] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> CrossEntropyWithSoftmax = CrossEntropyWithSoftmax (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalErrorPrediction = ErrorPrediction (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> EvalClassificationError = ClassificationError (labels, HLast) : [132 x *], [132 x 1 x *] -> [1]
|
||||
MPI Rank 2: Validating --> PosteriorProb = Softmax (HLast) : [132 x 1 x *] -> [132 x 1 x *]
|
||||
MPI Rank 2: Validating --> Prior = Mean (labels) : [132 x *] -> [132]
|
||||
MPI Rank 2: Validating --> LogOfPrior = Log (Prior) : [132] -> [132]
|
||||
|
|
@ -1756,7 +1756,7 @@ MPI Rank 2: 08/16/2016 03:03:41: Training criterion node(s):
|
|||
MPI Rank 2: 08/16/2016 03:03:41: CrossEntropyWithSoftmax = CrossEntropyWithSoftmax
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:41: Evaluation criterion node(s):
|
||||
MPI Rank 2: 08/16/2016 03:03:41: EvalErrorPrediction = ErrorPrediction
|
||||
MPI Rank 2: 08/16/2016 03:03:41: EvalClassificationError = ClassificationError
|
||||
MPI Rank 2:
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: Allocating matrices for forward and/or backward propagation.
|
||||
|
|
@ -1809,39 +1809,39 @@ MPI Rank 2: 08/16/2016 03:03:44: Starting Epoch 1: learning rate per sample = 0.
|
|||
MPI Rank 2: minibatchiterator: epoch 0: frames [0..20480] (first utterance at frame 0), data subset 0 of 1, with 1 datapasses
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:44: Starting minibatch loop.
|
||||
MPI Rank 2: 08/16/2016 03:03:44: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalErrorPrediction = 0.91718750 * 640; time = 0.2141s; samplesPerSecond = 2989.1
|
||||
MPI Rank 2: 08/16/2016 03:03:44: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalErrorPrediction = 0.92812500 * 640; time = 0.2062s; samplesPerSecond = 3103.9
|
||||
MPI Rank 2: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalErrorPrediction = 0.87343750 * 640; time = 0.2050s; samplesPerSecond = 3121.2
|
||||
MPI Rank 2: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalErrorPrediction = 0.84531250 * 640; time = 0.2064s; samplesPerSecond = 3100.1
|
||||
MPI Rank 2: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalErrorPrediction = 0.86406250 * 640; time = 0.2062s; samplesPerSecond = 3103.4
|
||||
MPI Rank 2: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalErrorPrediction = 0.86093750 * 640; time = 0.2078s; samplesPerSecond = 3079.6
|
||||
MPI Rank 2: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalErrorPrediction = 0.77812500 * 640; time = 0.2065s; samplesPerSecond = 3099.4
|
||||
MPI Rank 2: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalErrorPrediction = 0.83750000 * 640; time = 0.2062s; samplesPerSecond = 3103.9
|
||||
MPI Rank 2: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalErrorPrediction = 0.81250000 * 640; time = 0.2069s; samplesPerSecond = 3093.4
|
||||
MPI Rank 2: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalErrorPrediction = 0.80156250 * 640; time = 0.2055s; samplesPerSecond = 3115.1
|
||||
MPI Rank 2: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalErrorPrediction = 0.82187500 * 640; time = 0.2045s; samplesPerSecond = 3129.0
|
||||
MPI Rank 2: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalErrorPrediction = 0.77031250 * 640; time = 0.2062s; samplesPerSecond = 3104.3
|
||||
MPI Rank 2: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalErrorPrediction = 0.76093750 * 640; time = 0.2378s; samplesPerSecond = 2691.1
|
||||
MPI Rank 2: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalErrorPrediction = 0.73906250 * 640; time = 0.2061s; samplesPerSecond = 3104.6
|
||||
MPI Rank 2: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalErrorPrediction = 0.71093750 * 640; time = 0.2059s; samplesPerSecond = 3108.3
|
||||
MPI Rank 2: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalErrorPrediction = 0.74375000 * 640; time = 0.2048s; samplesPerSecond = 3124.4
|
||||
MPI Rank 2: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalErrorPrediction = 0.69531250 * 640; time = 0.2059s; samplesPerSecond = 3108.7
|
||||
MPI Rank 2: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalErrorPrediction = 0.64531250 * 640; time = 0.2052s; samplesPerSecond = 3118.9
|
||||
MPI Rank 2: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalErrorPrediction = 0.69843750 * 640; time = 0.2071s; samplesPerSecond = 3090.2
|
||||
MPI Rank 2: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalErrorPrediction = 0.68125000 * 640; time = 0.2058s; samplesPerSecond = 3109.3
|
||||
MPI Rank 2: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalErrorPrediction = 0.65312500 * 640; time = 0.2066s; samplesPerSecond = 3098.5
|
||||
MPI Rank 2: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalErrorPrediction = 0.63906250 * 640; time = 0.2069s; samplesPerSecond = 3093.4
|
||||
MPI Rank 2: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalErrorPrediction = 0.65000000 * 640; time = 0.2065s; samplesPerSecond = 3099.0
|
||||
MPI Rank 2: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalErrorPrediction = 0.65468750 * 640; time = 0.2057s; samplesPerSecond = 3111.1
|
||||
MPI Rank 2: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalErrorPrediction = 0.62812500 * 640; time = 0.2052s; samplesPerSecond = 3118.5
|
||||
MPI Rank 2: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalErrorPrediction = 0.62031250 * 640; time = 0.2050s; samplesPerSecond = 3122.1
|
||||
MPI Rank 2: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.2069s; samplesPerSecond = 3093.7
|
||||
MPI Rank 2: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalErrorPrediction = 0.66093750 * 640; time = 0.2056s; samplesPerSecond = 3113.2
|
||||
MPI Rank 2: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalErrorPrediction = 0.58906250 * 640; time = 0.2076s; samplesPerSecond = 3082.3
|
||||
MPI Rank 2: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalErrorPrediction = 0.60312500 * 640; time = 0.2103s; samplesPerSecond = 3044.0
|
||||
MPI Rank 2: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalErrorPrediction = 0.58125000 * 640; time = 0.2074s; samplesPerSecond = 3085.8
|
||||
MPI Rank 2: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalErrorPrediction = 0.59843750 * 640; time = 0.2053s; samplesPerSecond = 3116.8
|
||||
MPI Rank 2: 08/16/2016 03:03:51: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalErrorPrediction = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=6.65503s
|
||||
MPI Rank 2: 08/16/2016 03:03:44: Epoch[ 1 of 4]-Minibatch[ 1- 10, 3.13%]: CrossEntropyWithSoftmax = 4.56731190 * 640; EvalClassificationError = 0.91718750 * 640; time = 0.2141s; samplesPerSecond = 2989.1
|
||||
MPI Rank 2: 08/16/2016 03:03:44: Epoch[ 1 of 4]-Minibatch[ 11- 20, 6.25%]: CrossEntropyWithSoftmax = 4.31208878 * 640; EvalClassificationError = 0.92812500 * 640; time = 0.2062s; samplesPerSecond = 3103.9
|
||||
MPI Rank 2: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 21- 30, 9.38%]: CrossEntropyWithSoftmax = 3.97319840 * 640; EvalClassificationError = 0.87343750 * 640; time = 0.2050s; samplesPerSecond = 3121.2
|
||||
MPI Rank 2: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 31- 40, 12.50%]: CrossEntropyWithSoftmax = 3.73308124 * 640; EvalClassificationError = 0.84531250 * 640; time = 0.2064s; samplesPerSecond = 3100.1
|
||||
MPI Rank 2: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 41- 50, 15.63%]: CrossEntropyWithSoftmax = 3.83238242 * 640; EvalClassificationError = 0.86406250 * 640; time = 0.2062s; samplesPerSecond = 3103.4
|
||||
MPI Rank 2: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 51- 60, 18.75%]: CrossEntropyWithSoftmax = 3.69914238 * 640; EvalClassificationError = 0.86093750 * 640; time = 0.2078s; samplesPerSecond = 3079.6
|
||||
MPI Rank 2: 08/16/2016 03:03:45: Epoch[ 1 of 4]-Minibatch[ 61- 70, 21.88%]: CrossEntropyWithSoftmax = 3.40238588 * 640; EvalClassificationError = 0.77812500 * 640; time = 0.2065s; samplesPerSecond = 3099.4
|
||||
MPI Rank 2: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 71- 80, 25.00%]: CrossEntropyWithSoftmax = 3.51740313 * 640; EvalClassificationError = 0.83750000 * 640; time = 0.2062s; samplesPerSecond = 3103.9
|
||||
MPI Rank 2: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 81- 90, 28.13%]: CrossEntropyWithSoftmax = 3.50059778 * 640; EvalClassificationError = 0.81250000 * 640; time = 0.2069s; samplesPerSecond = 3093.4
|
||||
MPI Rank 2: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 91- 100, 31.25%]: CrossEntropyWithSoftmax = 3.39301549 * 640; EvalClassificationError = 0.80156250 * 640; time = 0.2055s; samplesPerSecond = 3115.1
|
||||
MPI Rank 2: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 101- 110, 34.38%]: CrossEntropyWithSoftmax = 3.48832144 * 640; EvalClassificationError = 0.82187500 * 640; time = 0.2045s; samplesPerSecond = 3129.0
|
||||
MPI Rank 2: 08/16/2016 03:03:46: Epoch[ 1 of 4]-Minibatch[ 111- 120, 37.50%]: CrossEntropyWithSoftmax = 3.23814723 * 640; EvalClassificationError = 0.77031250 * 640; time = 0.2062s; samplesPerSecond = 3104.3
|
||||
MPI Rank 2: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 121- 130, 40.63%]: CrossEntropyWithSoftmax = 3.14333583 * 640; EvalClassificationError = 0.76093750 * 640; time = 0.2378s; samplesPerSecond = 2691.1
|
||||
MPI Rank 2: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 131- 140, 43.75%]: CrossEntropyWithSoftmax = 3.01547841 * 640; EvalClassificationError = 0.73906250 * 640; time = 0.2061s; samplesPerSecond = 3104.6
|
||||
MPI Rank 2: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 141- 150, 46.88%]: CrossEntropyWithSoftmax = 2.91114805 * 640; EvalClassificationError = 0.71093750 * 640; time = 0.2059s; samplesPerSecond = 3108.3
|
||||
MPI Rank 2: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 151- 160, 50.00%]: CrossEntropyWithSoftmax = 3.06450741 * 640; EvalClassificationError = 0.74375000 * 640; time = 0.2048s; samplesPerSecond = 3124.4
|
||||
MPI Rank 2: 08/16/2016 03:03:47: Epoch[ 1 of 4]-Minibatch[ 161- 170, 53.13%]: CrossEntropyWithSoftmax = 2.77009796 * 640; EvalClassificationError = 0.69531250 * 640; time = 0.2059s; samplesPerSecond = 3108.7
|
||||
MPI Rank 2: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 171- 180, 56.25%]: CrossEntropyWithSoftmax = 2.67234909 * 640; EvalClassificationError = 0.64531250 * 640; time = 0.2052s; samplesPerSecond = 3118.9
|
||||
MPI Rank 2: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 181- 190, 59.38%]: CrossEntropyWithSoftmax = 2.76324613 * 640; EvalClassificationError = 0.69843750 * 640; time = 0.2071s; samplesPerSecond = 3090.2
|
||||
MPI Rank 2: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 191- 200, 62.50%]: CrossEntropyWithSoftmax = 2.70050608 * 640; EvalClassificationError = 0.68125000 * 640; time = 0.2058s; samplesPerSecond = 3109.3
|
||||
MPI Rank 2: 08/16/2016 03:03:48: Epoch[ 1 of 4]-Minibatch[ 201- 210, 65.63%]: CrossEntropyWithSoftmax = 2.56019594 * 640; EvalClassificationError = 0.65312500 * 640; time = 0.2066s; samplesPerSecond = 3098.5
|
||||
MPI Rank 2: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 211- 220, 68.75%]: CrossEntropyWithSoftmax = 2.56796356 * 640; EvalClassificationError = 0.63906250 * 640; time = 0.2069s; samplesPerSecond = 3093.4
|
||||
MPI Rank 2: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 221- 230, 71.88%]: CrossEntropyWithSoftmax = 2.51054929 * 640; EvalClassificationError = 0.65000000 * 640; time = 0.2065s; samplesPerSecond = 3099.0
|
||||
MPI Rank 2: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 231- 240, 75.00%]: CrossEntropyWithSoftmax = 2.52174700 * 640; EvalClassificationError = 0.65468750 * 640; time = 0.2057s; samplesPerSecond = 3111.1
|
||||
MPI Rank 2: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 241- 250, 78.13%]: CrossEntropyWithSoftmax = 2.45943503 * 640; EvalClassificationError = 0.62812500 * 640; time = 0.2052s; samplesPerSecond = 3118.5
|
||||
MPI Rank 2: 08/16/2016 03:03:49: Epoch[ 1 of 4]-Minibatch[ 251- 260, 81.25%]: CrossEntropyWithSoftmax = 2.36070476 * 640; EvalClassificationError = 0.62031250 * 640; time = 0.2050s; samplesPerSecond = 3122.1
|
||||
MPI Rank 2: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 261- 270, 84.38%]: CrossEntropyWithSoftmax = 2.22167676 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.2069s; samplesPerSecond = 3093.7
|
||||
MPI Rank 2: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 271- 280, 87.50%]: CrossEntropyWithSoftmax = 2.48104909 * 640; EvalClassificationError = 0.66093750 * 640; time = 0.2056s; samplesPerSecond = 3113.2
|
||||
MPI Rank 2: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 281- 290, 90.63%]: CrossEntropyWithSoftmax = 2.23253572 * 640; EvalClassificationError = 0.58906250 * 640; time = 0.2076s; samplesPerSecond = 3082.3
|
||||
MPI Rank 2: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 291- 300, 93.75%]: CrossEntropyWithSoftmax = 2.22145425 * 640; EvalClassificationError = 0.60312500 * 640; time = 0.2103s; samplesPerSecond = 3044.0
|
||||
MPI Rank 2: 08/16/2016 03:03:50: Epoch[ 1 of 4]-Minibatch[ 301- 310, 96.88%]: CrossEntropyWithSoftmax = 2.21771892 * 640; EvalClassificationError = 0.58125000 * 640; time = 0.2074s; samplesPerSecond = 3085.8
|
||||
MPI Rank 2: 08/16/2016 03:03:51: Epoch[ 1 of 4]-Minibatch[ 311- 320, 100.00%]: CrossEntropyWithSoftmax = 2.19995645 * 640; EvalClassificationError = 0.59843750 * 640; time = 0.2053s; samplesPerSecond = 3116.8
|
||||
MPI Rank 2: 08/16/2016 03:03:51: Finished Epoch[ 1 of 4]: [Training] CrossEntropyWithSoftmax = 3.00789787 * 20480; EvalClassificationError = 0.72641602 * 20480; totalSamplesSeen = 20480; learningRatePerSample = 0.015625; epochTime=6.65503s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:54: Starting Epoch 2: learning rate per sample = 0.001953 effective momentum = 0.656119 momentum as time constant = 607.5 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 1: frames [20480..40960] (first utterance at frame 20480), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1850,45 +1850,45 @@ MPI Rank 2: 08/16/2016 03:03:54: Starting minibatch loop, DataParallelSGD traini
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.040872
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.04797
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.070448
|
||||
MPI Rank 2: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.14273439 * 2304; EvalErrorPrediction = 0.57074653 * 2304; time = 0.6233s; samplesPerSecond = 3696.6
|
||||
MPI Rank 2: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 1- 10, 12.50%]: CrossEntropyWithSoftmax = 2.14273439 * 2304; EvalClassificationError = 0.57074653 * 2304; time = 0.6233s; samplesPerSecond = 3696.6
|
||||
MPI Rank 2: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.048685
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.044716
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.044362
|
||||
MPI Rank 2: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.16748503 * 2560; EvalErrorPrediction = 0.58085937 * 2560; time = 0.5372s; samplesPerSecond = 4765.8
|
||||
MPI Rank 2: 08/16/2016 03:03:55: Epoch[ 2 of 4]-Minibatch[ 11- 20, 25.00%]: CrossEntropyWithSoftmax = 2.16748503 * 2560; EvalClassificationError = 0.58085937 * 2560; time = 0.5372s; samplesPerSecond = 4765.8
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.007114
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.040694
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.000949
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.071457
|
||||
MPI Rank 2: 08/16/2016 03:03:56: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.20296428 * 2560; EvalErrorPrediction = 0.58945313 * 2560; time = 0.5867s; samplesPerSecond = 4363.4
|
||||
MPI Rank 2: 08/16/2016 03:03:56: Epoch[ 2 of 4]-Minibatch[ 21- 30, 37.50%]: CrossEntropyWithSoftmax = 2.20296428 * 2560; EvalClassificationError = 0.58945313 * 2560; time = 0.5867s; samplesPerSecond = 4363.4
|
||||
MPI Rank 2: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.006542
|
||||
MPI Rank 2: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.030428
|
||||
MPI Rank 2: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.19897771 * 2560; EvalErrorPrediction = 0.58867187 * 2560; time = 0.5313s; samplesPerSecond = 4818.4
|
||||
MPI Rank 2: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 31- 40, 50.00%]: CrossEntropyWithSoftmax = 2.19897771 * 2560; EvalClassificationError = 0.58867187 * 2560; time = 0.5313s; samplesPerSecond = 4818.4
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.041416
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.006786
|
||||
MPI Rank 2: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.0563
|
||||
MPI Rank 2: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.13999099 * 2560; EvalErrorPrediction = 0.58320313 * 2560; time = 0.5817s; samplesPerSecond = 4401.1
|
||||
MPI Rank 2: 08/16/2016 03:03:57: Epoch[ 2 of 4]-Minibatch[ 41- 50, 62.50%]: CrossEntropyWithSoftmax = 2.13999099 * 2560; EvalClassificationError = 0.58320313 * 2560; time = 0.5817s; samplesPerSecond = 4401.1
|
||||
MPI Rank 2: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.076147
|
||||
MPI Rank 2: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.044663
|
||||
MPI Rank 2: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10194078 * 2560; EvalErrorPrediction = 0.56875000 * 2560; time = 0.5856s; samplesPerSecond = 4371.5
|
||||
MPI Rank 2: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 51- 60, 75.00%]: CrossEntropyWithSoftmax = 2.10194078 * 2560; EvalClassificationError = 0.56875000 * 2560; time = 0.5856s; samplesPerSecond = 4371.5
|
||||
MPI Rank 2: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.056967
|
||||
MPI Rank 2: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.006004
|
||||
MPI Rank 2: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10371395 * 2560; EvalErrorPrediction = 0.56328125 * 2560; time = 0.5531s; samplesPerSecond = 4628.5
|
||||
MPI Rank 2: 08/16/2016 03:03:58: Epoch[ 2 of 4]-Minibatch[ 61- 70, 87.50%]: CrossEntropyWithSoftmax = 2.10371395 * 2560; EvalClassificationError = 0.56328125 * 2560; time = 0.5531s; samplesPerSecond = 4628.5
|
||||
MPI Rank 2: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.054764
|
||||
MPI Rank 2: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.053099
|
||||
MPI Rank 2: 08/16/2016 03:03:59: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.08426773 * 2560; EvalErrorPrediction = 0.58398438 * 2560; time = 0.6065s; samplesPerSecond = 4220.7
|
||||
MPI Rank 2: 08/16/2016 03:03:59: Epoch[ 2 of 4]-Minibatch[ 71- 80, 100.00%]: CrossEntropyWithSoftmax = 2.08426773 * 2560; EvalClassificationError = 0.58398438 * 2560; time = 0.6065s; samplesPerSecond = 4220.7
|
||||
MPI Rank 2: Async gradient aggregation wait time: 1e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.025261
|
||||
MPI Rank 2: 08/16/2016 03:03:59: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14176518 * 20480; EvalErrorPrediction = 0.57827148 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.63717s
|
||||
MPI Rank 2: 08/16/2016 03:03:59: Finished Epoch[ 2 of 4]: [Training] CrossEntropyWithSoftmax = 2.14176518 * 20480; EvalClassificationError = 0.57827148 * 20480; totalSamplesSeen = 40960; learningRatePerSample = 0.001953125; epochTime=4.63717s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:03:59: Starting Epoch 3: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 2: frames [40960..61440] (first utterance at frame 40960), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1898,13 +1898,13 @@ MPI Rank 2: Async gradient aggregation wait time: 0.103418
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.168332
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.014615
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.208225
|
||||
MPI Rank 2: 08/16/2016 03:04:00: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11704698 * 9216; EvalErrorPrediction = 0.55631510 * 9216; time = 1.4484s; samplesPerSecond = 6362.9
|
||||
MPI Rank 2: 08/16/2016 03:04:00: Epoch[ 3 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 2.11704698 * 9216; EvalClassificationError = 0.55631510 * 9216; time = 1.4484s; samplesPerSecond = 6362.9
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.087413
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.16582
|
||||
MPI Rank 2: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.007639
|
||||
MPI Rank 2: 08/16/2016 03:04:02: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97705226 * 10240; EvalErrorPrediction = 0.54746094 * 10240; time = 1.5635s; samplesPerSecond = 6549.4
|
||||
MPI Rank 2: 08/16/2016 03:04:02: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.04081650 * 20480; EvalErrorPrediction = 0.55122070 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=3.06497s
|
||||
MPI Rank 2: 08/16/2016 03:04:02: Epoch[ 3 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.97705226 * 10240; EvalClassificationError = 0.54746094 * 10240; time = 1.5635s; samplesPerSecond = 6549.4
|
||||
MPI Rank 2: 08/16/2016 03:04:02: Finished Epoch[ 3 of 4]: [Training] CrossEntropyWithSoftmax = 2.04081650 * 20480; EvalClassificationError = 0.55122070 * 20480; totalSamplesSeen = 61440; learningRatePerSample = 9.7656251e-005; epochTime=3.06497s
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:04:02: Starting Epoch 4: learning rate per sample = 0.000098 effective momentum = 0.656119 momentum as time constant = 2429.9 samples
|
||||
MPI Rank 2: minibatchiterator: epoch 3: frames [61440..81920] (first utterance at frame 61440), data subset 2 of 3, with 1 datapasses
|
||||
|
|
@ -1914,14 +1914,14 @@ MPI Rank 2: Async gradient aggregation wait time: 3e-006
|
|||
MPI Rank 2: Actual gradient aggregation time: 0.092817
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.095403
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.083889
|
||||
MPI Rank 2: 08/16/2016 03:04:04: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.92306138 * 9216; EvalErrorPrediction = 0.52398003 * 9216; time = 1.3936s; samplesPerSecond = 6613.1
|
||||
MPI Rank 2: 08/16/2016 03:04:04: Epoch[ 4 of 4]-Minibatch[ 1- 10, 50.00%]: CrossEntropyWithSoftmax = 1.92306138 * 9216; EvalClassificationError = 0.52398003 * 9216; time = 1.3936s; samplesPerSecond = 6613.1
|
||||
MPI Rank 2: Async gradient aggregation wait time: 3e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.00513
|
||||
MPI Rank 2: Async gradient aggregation wait time: 2e-006
|
||||
MPI Rank 2: Actual gradient aggregation time: 0.006516
|
||||
MPI Rank 2: 08/16/2016 03:04:05: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91800591 * 10240; EvalErrorPrediction = 0.52324219 * 10240; time = 1.5838s; samplesPerSecond = 6465.3
|
||||
MPI Rank 2: 08/16/2016 03:04:05: Epoch[ 4 of 4]-Minibatch[ 11- 20, 100.00%]: CrossEntropyWithSoftmax = 1.91800591 * 10240; EvalClassificationError = 0.52324219 * 10240; time = 1.5838s; samplesPerSecond = 6465.3
|
||||
MPI Rank 2: Async gradient aggregation wait time: 0.001211
|
||||
MPI Rank 2: 08/16/2016 03:04:05: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.92214429 * 20480; EvalErrorPrediction = 0.52407227 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=3.0999s
|
||||
MPI Rank 2: 08/16/2016 03:04:05: Finished Epoch[ 4 of 4]: [Training] CrossEntropyWithSoftmax = 1.92214429 * 20480; EvalClassificationError = 0.52407227 * 20480; totalSamplesSeen = 81920; learningRatePerSample = 9.7656251e-005; epochTime=3.0999s
|
||||
MPI Rank 2: 08/16/2016 03:04:05: CNTKCommandTrainEnd: speechTrain
|
||||
MPI Rank 2:
|
||||
MPI Rank 2: 08/16/2016 03:04:05: Action "train" complete.
|
||||
|
|
|
|||
Некоторые файлы не были показаны из-за слишком большого количества измененных файлов Показать больше
Загрузка…
Ссылка в новой задаче