More ont_hot fixes.

Examples/Image/Classification/ConvNet/Python/ConvNet_CIFAR10_DataAug.py

@@ -115,7 +115,7 @@ def train_and_evaluate(reader, reader_test, model, epoch_size=50000, max_epochs=
     criterion = create_criterion_function(model, normalize=lambda x: x / 256)
     #debughelpers.dump_function(criterion, 'criterion')
 
-    #from cntk.graph import plot
+    #from cntk.logging.graph import plot
     #plot(criterion, filename=os.path.join(model_path, "ConvNet_CIFAR10_DataAug.pdf"))
 
     # iteration parameters

Examples/LanguageUnderstanding/ATIS/Python/LanguageUnderstanding.py

@@ -137,7 +137,7 @@ def train(reader, model, max_epochs):
     labels = reader.streams.slot_labels
     #labels = reader.streams.intent_labels  # for intent classification
 
-    #from cntk.graph import plot
+    #from cntk.logging.graph import plot
     #plot(criterion, filename=data_dir + "/model.pdf")
 
     # iteration parameters  --needed here because learner schedule needs it

Examples/Text/WordLMWithSampledSoftmax/data_reader.py

@@ -78,7 +78,7 @@ class DataReader(object):
 
             # From the end of the file there are probably some leftover lines
             if len(feature_sequences) > 0:
-                yield C.one_hot(feature_sequences, self.vocab_dim), C.one_hot(label_sequences, self.vocab_dim), token_count
+                yield C.Value.one_hot(feature_sequences, self.vocab_dim), C.one_hot(label_sequences, self.vocab_dim), token_count
 
 
 

bindings/python/cntk/core.py

@@ -375,7 +375,7 @@ class Value(cntk_py.Value):
             >>> sparse_indices = [[1,5],[4]]
             >>> i0 = C.input_variable(shape=num_classes, is_sparse=True)
             >>> z = C.times(i0, np.eye(num_classes))
-            >>> value = C.one_hot(sparse_indices, num_classes)
+            >>> value = C.Value.one_hot(sparse_indices, num_classes)
             >>> z.eval({i0: value})
             [array([[ 0.,  1.,  0.,  0.,  0.,  0.],
                     [ 0.,  0.,  0.,  0.,  0.,  1.]], dtype=float32), 
@@ -386,7 +386,7 @@ class Value(cntk_py.Value):
             >>> sparse_indices = [[1,5,3,2],[4,1]]
             >>> i0 = C.input_variable(shape=sample_shape, is_sparse=True)
             >>> z = C.times(i0, np.eye(num_classes))
-            >>> value = C.one_hot(sparse_indices, sample_shape)
+            >>> value = C.Value.one_hot(sparse_indices, sample_shape)
             >>> z.eval({i0: value})
             [array([[[ 0.,  1.,  0.,  0.,  0.,  0.],
                      [ 0.,  0.,  0.,  0.,  0.,  1.]],

bindings/python/cntk/debugging/debug.py

@@ -368,7 +368,7 @@ class _DebugNode(UserFunction):
 
 
 def _nodes_to_debug(model):
-    from cntk.graph import depth_first_search
+    from cntk.logging.graph import depth_first_search
 
     def node_filter(x):
         if hasattr(x, 'op_name') and x.op_name in ['NoOp']:
@@ -421,7 +421,7 @@ def debug_model(model, in_stream=sys.stdin, out_stream=sys.stdout,
                          for n in nodes}
 
         model = model.clone(CloneMethod.share, modifications)
-        from cntk.graph import plot
+        from cntk.logging.graph import plot
 
         nodes = _nodes_to_debug(model)
         if len(nodes) == 1:

bindings/python/cntk/layers/layers.py

@@ -158,7 +158,7 @@ def Embedding(shape=None, init=default_override_or(glorot_uniform()), weights=No
      >>> e = f(x)
      >>> e.shape
          (5,)
-     >>> e(C.one_hot([[1], [0], [0], [1]], num_classes=2))
+     >>> e(C.Value.one_hot([[1], [0], [0], [1]], num_classes=2))
      array([[[ 0.7,  0.6,  0.3,  0.2,  0.9]],
      <BLANKLINE>
             [[ 0.5,  0.3,  0.1,  0.4,  0.2]],

bindings/python/cntk/ops/__init__.py

@@ -2655,19 +2655,6 @@ def constant(value=None, shape=None, dtype=None, device=None, name=''):
         :class:`~cntk.ops.variables.Constant`
     '''
     from .variables import Constant
-    if not device:
-        device = use_default_device()
-    if (np.isscalar(value) or isinstance(value, np.ndarray)) and not shape:
-        shape = ()
-    if dtype is not None:
-        if isinstance(value, np.ndarray) and dtype != value.dtype:
-            value = np.array(value, dtype=dtype)
-    else:
-        if isinstance(value, np.ndarray):
-            dtype = value.dtype
-        else:
-            dtype = np.float32
-
     return Constant(value, shape, dtype, device, name)
 
 ##########################################################################

bindings/python/cntk/ops/functions.py

@@ -542,7 +542,7 @@ class Function(cntk_py.Function):
             >>> # 1st sequence: word 1
             >>> # 2nd sequence: words 2 and 4
             >>> batch = [[1],[2,4]]
-            >>> sparse_batch = C.one_hot(batch, vocab_size)
+            >>> sparse_batch = C.Value.one_hot(batch, vocab_size)
             >>> _, fv = f.forward({v:sparse_batch})
             >>> list(fv.values())[0]
             [array([[ 0.,  1.,  0.,  0.,  0.]], dtype=float32),

bindings/python/cntk/ops/variables.py

@@ -300,15 +300,21 @@ class Constant(VariableMixin, TensorOpsMixin, cntk_py.Constant):
     '''
     def __init__(self, value=None, shape=None, dtype=None, device=None, name=''):
 
-        if dtype is None:
+        if not device:
+            device = use_default_device()
+
+        if (np.isscalar(value) or isinstance(value, np.ndarray)) and not shape:
+            shape = ()
+
+        if dtype is not None:
+            if isinstance(value, np.ndarray) and dtype != value.dtype:
+                value = np.array(value, dtype=dtype)
+        else:
             if isinstance(value, np.ndarray):
                 dtype = value.dtype
             else:
                 dtype = np.float32
 
-        if device is None:
-            device = DeviceDescriptor.use_default_device()
-
         if np.isscalar(value):
             super(Constant, self).__init__(sanitize_shape(shape),
                     sanitize_dtype_cntk(dtype), value, device, name)