bed7c19cfb
Integrate zhouwang/evaldll-client-vs2015 into master
2017-01-12 07:37:49 -08:00
2a4c7f8031
fix tabs and add a wiki link
2017-01-12 12:31:06 +01:00
93b2133dc4
change Readme file too.
2017-01-12 11:15:16 +01:00
35bcd6365c
upgrade the solution file for eval samples to VS2015, remove EvalV2clients (it will be in another solution for V2
2017-01-12 10:19:42 +01:00
9f894adcf1
Integrate kedeng/BNCuDnn2 into master
2017-01-11 13:46:16 -08:00
a4e94eb1ff
Moving feature extraction example to ResNet_18
2017-01-11 12:48:43 +01:00
fc7331ae6f
added test for feature extraction example
2017-01-11 12:48:43 +01:00
319386f528
addressed CR comments
2017-01-11 12:48:43 +01:00
2bd3cc8f22
Feature extraction example using AlexNet from python
2017-01-11 12:48:42 +01:00
6c73818bf7
[CNTK Core] Change BatchNormalization default to CuDnn in both BrainScript and V2 C++/Python, and add fallback to cntk engine when not supported by cudnn
...
Fix epsilon in CuDnnBatchNormalization to be consistent in forward/backward
Fix BatchNormalization parameter gradient accumulation
Add message on CuDnnBatchNormalization forward infer batch size limitation, and adjust CNTK_202_Language_Understanding eval minibatch size to avoid CUDA failure
2017-01-11 01:00:09 -08:00
434ca582e7
Examples/Image/Classification/ResNet/BrainScript: Linux: point out that stack size may need to be increased
...
Closes : #1273
References: #1074
2017-01-11 08:48:51 +01:00
7cb1db6620
Merge remote-tracking branch 'origin/master' into mahilleb/vs2015fi
...
Conflicts:
Source/CNTKv2LibraryDll/Utils.cpp
2017-01-10 11:29:03 +01:00
911b2ca82e
Windows: switch to Visual Studio 2015
2017-01-10 10:47:35 +01:00
c9b03e49b3
Integrate mahilleb/Mnist13 into master
2017-01-10 01:14:03 -08:00
eda6561105
Remove unnecesarry new line that was mistakenly added
2017-01-10 10:00:00 +02:00
1cbc4ed9c6
Fixed an issue with incorrect bounding boxes read when gt annotation contains only one regio
2017-01-10 09:58:09 +02:00
4b31eda24f
Merge remote-tracking branch 'origin/master' into mahilleb/Mnist13
2017-01-10 08:05:44 +01:00
6dc09565bb
CNTK_103B_MNIST_FeedForwardNetwork.ipynb: add test
...
Also, let run MNIST tests against the default (end-user) data location.
2017-01-10 05:42:30 +01:00
a51e7fcab2
Update cntk file.
2017-01-09 15:24:22 -08:00
5674d33221
update binding based on the latest API, update samples, add buidling vocabulary index in samples for the atis model.
2017-01-07 10:29:58 +01:00
1b122c484e
rename CNTKLibraryManagedClient to CNTKLibraryManagedExamples
2017-01-07 10:29:57 +01:00
a0554dfe9e
improve samples
2017-01-07 10:29:57 +01:00
4b9cc9c040
add CreateSequence without SeqStartFlag
2017-01-07 10:29:57 +01:00
0b47f30ad0
remove check for PackedValue as it will be automatically unpacked when accessing Data/Mask; fix bugs
2017-01-07 10:29:57 +01:00
f7a2172828
move CSEvalV2Examples to the examples folder and move the project files to the Test folder
2017-01-07 10:29:57 +01:00
570a765887
Revert "Add ReasoNet model as example"
...
This reverts commit 09daade4f0
2017-01-06 22:58:44 +08:00
77e5d159ce
Revert "Add loss/pred for ReasoNet"
...
This reverts commit 2f94c5430b
2017-01-06 22:58:44 +08:00
d1bed651f0
Revert "Add training method for ReasoNet"
...
This reverts commit 732836ab00
2017-01-06 22:58:44 +08:00
510595348f
Revert "Fix training issues with ReasoNet"
...
This reverts commit 1e0f8629ca
2017-01-06 22:58:44 +08:00
1e0f8629ca
Fix training issues with ReasoNet
2017-01-06 22:34:39 +08:00
732836ab00
Add training method for ReasoNet
2017-01-06 22:34:39 +08:00
2f94c5430b
Add loss/pred for ReasoNet
2017-01-06 22:34:39 +08:00
09daade4f0
Add ReasoNet model as example
2017-01-06 22:34:39 +08:00
caee1a5d9b
Merge pull request #1217 from miguelgfierro/patch-1
...
Fixed bug in ConvNet_MNIST.py
2017-01-05 08:50:32 -08:00
9862bf34fc
Merge remote-tracking branch 'origin/pr/1182' into mahilleb/SampleFix
2017-01-05 13:41:46 +01:00
1abd7e9fcd
Fixed bug in ConvNet_MNIST.py
...
When running `ConvNet_MNIST.py` I get the following error:
```bash
Traceback (most recent call last):
File "ConvNet_MNIST.py", line 131, in <module>
convnet_mnist()
File "ConvNet_MNIST.py", line 117, in convnet_mnist
metric_numer += trainer.test_minibatch(data) * current_minibatch
File "/home/hoaphumanoid/anaconda3/envs/cntk-py34/lib/python3.4/site-packages/cntk/trainer.py", line 125, in test_minibatch
arguments = sanitize_var_map(self.model.arguments, arguments)
File "/home/hoaphumanoid/anaconda3/envs/cntk-py34/lib/python3.4/site-packages/cntk/utils/__init__.py", line 373, in sanitize_var_map
len(op_arguments))
ValueError: function expects 1 arguments
```
2016-12-29 10:56:13 +00:00
970daa8d43
Added use of path.join to better accomodate windows scenarios
2016-12-25 10:03:23 +02:00
a01e609450
Integrate nikosk/highlight into master
2016-12-22 17:28:57 -08:00
a8fa4554ee
Removed annoying ?highlight from pythondocs links
2016-12-22 16:41:28 -08:00
7f520dfe99
Update README.
2016-12-22 14:01:53 -08:00
0092a9d706
Update README.
2016-12-22 14:00:40 -08:00
89be1c4bf1
Merge branch 'master' of https://github.com/Microsoft/CNTK
2016-12-22 13:44:46 -08:00
35ff61d92d
Add readme files for GoogLeNet and AlexNet.
2016-12-22 13:44:12 -08:00
87dd18d95a
Integrate zhouwang/outputname into master
2016-12-22 11:31:40 -08:00
876ee17b52
remove _output suffix from name of output variables
...
add unit tests for output variable name; adapt tests to accomodate name change
restore _output suffix for uid; test function name only on CPU, as it is device neutral.
Correctly handle visited nodes after removing _output suffix
2016-12-22 17:25:30 +01:00
0849ac9110
Examples/SequenceClassification/SimpleExample/Python/SequenceClassification.py: fix sys.path.append()
2016-12-22 10:53:14 +01:00
0ff43614b7
Examples/Evaluation/CSEvalClient: bump NuGet package version
2016-12-22 10:53:01 +01:00
50f85dee66
Integrate wolfma/rand13bl-lin2 into master
2016-12-21 11:16:34 -08:00
21ef59cc91
Update AlexNet_ImageNet.cntk
...
Remove commented part.
2016-12-21 10:11:25 -08:00
e2fd7255af
Align randomization across platforms
2016-12-21 17:31:22 +01:00
Project Philly
Zhou Wang
Philipp Kranen
KeDengMS
Mark Hillebrand
Nadav Bar
Cha Zhang
PengchengHe
Miguel González-Fierro
Nadav Bar
Nikos Karampatziakis
Cha Zhang