621561a | Ivan Rodriguez | 20 June 2016, 09:51:37 UTC | check of original tests | 03 August 2016, 10:56:19 UTC |
f951e9a | Thilo Will | 03 August 2016, 09:35:59 UTC | Fixing typo in comment | 03 August 2016, 09:35:59 UTC |
0d3b9e5 | Thilo Will | 03 August 2016, 08:59:12 UTC | Added comment regarding default axis values in python bindings | 03 August 2016, 08:59:12 UTC |
01915bc | Zhou Wang | 29 July 2016, 16:58:49 UTC | Link libcntkmath to libeval Use GDK_NVML_LIB_PATH Simplify Boost libraries Use ORIGINDIR in rpath of libeval remove GDK_NVML_LIB_PATH from cppevalclient and Add -lcntkmath to libeval.so Use GDK_NVML_LIB_PATH to avoid using system-wide one, Simplify use in libeval.so | 03 August 2016, 06:43:47 UTC |
aaac609 | Frank Seide | 03 August 2016, 05:07:46 UTC | bug fix: LearnableParameter::CopyTo() must exist, now that we have the lazy-init variables | 03 August 2016, 05:07:46 UTC |
c213625 | Frank Seide | 03 August 2016, 04:22:07 UTC | NDL version of lazy Parameter init (for dim inference) | 03 August 2016, 04:22:07 UTC |
e80562f | Frank Seide | 03 August 2016, 02:24:42 UTC | trying a fix to lazy init | 03 August 2016, 02:24:42 UTC |
bc06c3c | Frank Seide | 03 August 2016, 02:13:30 UTC | CNTK BatchNorm engine Backprop() should honor blendFactor | 03 August 2016, 02:13:30 UTC |
838717f | Frank Seide | 02 August 2016, 21:51:56 UTC | bug fix in lazy random init of Parameter: Validate() is called too early, must do it in InferInputDimsFrom() | 02 August 2016, 21:51:56 UTC |
6232550 | Dong Yu | 02 August 2016, 21:11:38 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into dongyu/stoplr | 02 August 2016, 21:11:38 UTC |
0e3b5da | Zhou Wang | 02 August 2016, 16:35:08 UTC | include crtdbg.h only on Windows | 02 August 2016, 16:35:08 UTC |
f61cc4e | Thilo Will | 02 August 2016, 15:21:32 UTC | in __init__.py boved ReduceLogSum to bottom. Renamed python operation to reduce_log_sum | 02 August 2016, 15:21:32 UTC |
8db871e | Wolfgang Manousek | 02 August 2016, 15:08:12 UTC | addressed final code review comments | 02 August 2016, 15:09:54 UTC |
e440132 | Wolfgang Manousek | 02 August 2016, 14:44:36 UTC | moved to _CrtSetReportHook2 | 02 August 2016, 15:09:54 UTC |
e7fd489 | Wolfgang Manousek | 02 August 2016, 10:58:43 UTC | use CrtSetReportHook on windows to disable the popup of assert boxes | 02 August 2016, 15:09:54 UTC |
f84fdc2 | Thilo Will | 02 August 2016, 14:57:56 UTC | ReduceLogSum: testcases for backward path. Documentation | 02 August 2016, 14:57:56 UTC |
e2e15e0 | Project Philly | 02 August 2016, 13:53:24 UTC | Integrate mahilleb/AssertRemoval into master | 02 August 2016, 13:53:24 UTC |
fa7befb | Mark A. Hillebrand | 02 August 2016, 13:48:54 UTC | Address CR comment | 02 August 2016, 13:48:54 UTC |
6c7f7c1 | Thilo Will | 02 August 2016, 11:38:47 UTC | Forward path test for ReduceLogSum works | 02 August 2016, 11:38:47 UTC |
cf51d75 | Thilo Will | 02 August 2016, 07:43:30 UTC | Adding code for reduce_logsum test. | 02 August 2016, 07:43:30 UTC |
5a7e77b | thilow | 02 August 2016, 07:32:57 UTC | ElementwiseProductWithExpOffDiff | 02 August 2016, 07:32:57 UTC |
2115ac1 | Dong Yu | 01 August 2016, 22:30:57 UTC | allow to customize minLearningRatePerSample to control early stopping. | 01 August 2016, 22:30:57 UTC |
c81eb6f | Vadim Mazalov | 27 July 2016, 00:01:25 UTC | Add Array struct, quantizer unit tests, minor fixes. | 01 August 2016, 19:33:03 UTC |
7cfc3f3 | Vadim Mazalov | 27 June 2016, 16:23:56 UTC | Remove LearnableParameterQuantized and MEL command to quantize a node | 01 August 2016, 19:33:03 UTC |
90079c6 | Vadim Mazalov | 16 June 2016, 18:26:36 UTC | Introduce Matrix<short>. | 01 August 2016, 19:28:11 UTC |
15e9cf8 | Vadim Mazalov | 16 June 2016, 03:36:45 UTC | Refine MEL command for quantization of LearnableParameter node, changes to InputAndParamNodes. | 01 August 2016, 19:28:11 UTC |
77ff661 | Vadim Mazalov | 10 June 2016, 04:54:26 UTC | Quantization of learnable parameter node | 01 August 2016, 19:28:11 UTC |
a6d994b | Vadim Mazalov | 09 June 2016, 03:07:56 UTC | Add LearnableParameterQuantized and Quantizer. | 01 August 2016, 19:11:12 UTC |
97c0c98 | Mark A. Hillebrand | 01 August 2016, 16:46:09 UTC | Remove assertion that's not true for the CuDNN engine. | 01 August 2016, 16:46:09 UTC |
dc2c4f6 | Thilo Will | 01 August 2016, 12:35:42 UTC | Undoing some more previous changes | 01 August 2016, 12:35:42 UTC |
0d827e6 | Mark A. Hillebrand | 01 August 2016, 12:27:38 UTC | configure: fix detection of LibZip when path explicitly specified Closes: #713. | 01 August 2016, 12:27:38 UTC |
fd36106 | Thilo Will | 01 August 2016, 12:19:05 UTC | Undoing core.bs changes related to basing softmax on reducelogsum | 01 August 2016, 12:19:05 UTC |
0d34258 | Frank Seide | 31 July 2016, 06:34:51 UTC | bug fix in Image/QuickE2E() DNNSigmoidLayer() call when used with dim inference | 31 July 2016, 06:34:51 UTC |
b266f24 | Frank Seide | 31 July 2016, 06:11:51 UTC | (minor fixes of previous commit) | 31 July 2016, 06:11:51 UTC |
3be923a | Frank Seide | 31 July 2016, 05:56:50 UTC | bug fix: LearnableParameter now defers random initialization until after validation, to support inference of parameter dimensions during validation | 31 July 2016, 05:56:50 UTC |
ec4314e | Frank Seide | 31 July 2016, 04:55:56 UTC | refactored LSTM E2E test to separate out the model() function | 31 July 2016, 04:55:56 UTC |
87acc99 | Frank Seide | 31 July 2016, 04:41:01 UTC | Parameter() no longer requires to say init="...", saying initValue=... or initFromFilePath=... is enough | 31 July 2016, 04:41:01 UTC |
150ea69 | Frank Seide | 31 July 2016, 02:55:46 UTC | ConfigArray::GetIndexRange() changed to return [begin, end) (excluding end value) | 31 July 2016, 02:55:46 UTC |
abb6811 | Frank Seide | 31 July 2016, 02:49:19 UTC | implemented array-literal syntax { a, b, c } (test case: Speech/QuickE2E); empty arrays are now allowed at places; bug fix: unary operators should not have precedence higher than record lookup, array index, or function invocation | 31 July 2016, 02:49:19 UTC |
7e51e44 | Frank Seide | 31 July 2016, 02:25:01 UTC | added { } parentheses as an alternative in function definition/invocation, intended for model-creating constructors; adapted Speech QuickE2E test to this | 31 July 2016, 02:25:01 UTC |
0854708 | Amit Agarwal | 30 July 2016, 23:16:41 UTC | CNTK v2 library: a) Convolution b) Mean variance normalization c) Added Cifar resnet test | 30 July 2016, 23:16:41 UTC |
5c8f39c | Frank Seide | 30 July 2016, 04:53:27 UTC | refined QuickE2E sample further | 30 July 2016, 04:53:27 UTC |
1e8ad6f | Frank Seide | 30 July 2016, 04:13:37 UTC | Speech QuickE2E reimplemented in BrainScript with Keras-like layer functions | 30 July 2016, 04:13:37 UTC |
6a5e971 | Frank Seide | 30 July 2016, 02:34:44 UTC | Merge branch 'master' of https://github.com/Microsoft/cntk into fseide/clonebs | 30 July 2016, 02:34:44 UTC |
0192ba5 | Frank Seide | 30 July 2016, 02:31:35 UTC | new option 'logGradientToo' for TraceNode | 30 July 2016, 02:31:35 UTC |
fc5ccd7 | Frank Seide | 29 July 2016, 22:41:05 UTC | TrainOrAdaptModel() now logs the names of the nodes that are actually updated | 29 July 2016, 22:41:05 UTC |
965999b | U-FAREAST\fseide | 29 July 2016, 22:17:14 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into fseide/clonebs | 29 July 2016, 22:17:14 UTC |
6293392 | U-FAREAST\fseide | 29 July 2016, 22:17:06 UTC | strange change detected in cmudict-0.7b.mapping | 29 July 2016, 22:17:06 UTC |
0c86e36 | Frank Seide | 29 July 2016, 19:42:23 UTC | bug fix: BatchNormEngine::Forward() should assert saveMean/InvStdDev as a post-condition now | 29 July 2016, 19:42:23 UTC |
12b2c31 | Project Philly | 29 July 2016, 12:43:46 UTC | Integrate mahilleb/DockerNoGdk into master | 29 July 2016, 12:43:46 UTC |
a30f0af | Mark A. Hillebrand | 29 July 2016, 08:33:46 UTC | Windows: fix for creating build information In CNTK.vcxproj and prebuild.bat, do not rely on environment variables, but pass down MSBuild properties. Related: #723. | 29 July 2016, 11:48:46 UTC |
efd5cba | Mark Hillebrand | 27 July 2016, 14:26:58 UTC | Tools/docker/Dockerfile-GPU: use pre-installed GDK Closes #629. | 29 July 2016, 11:09:51 UTC |
018877f | Mark Hillebrand | 27 July 2016, 14:14:34 UTC | Linux: Makefile/configure: split GDK_PATH setting into two, include and NVML path | 29 July 2016, 11:09:27 UTC |
182b037 | Zhou Wang | 27 July 2016, 19:31:40 UTC | Use ORIGINLIBDIR instead of LIBDIR in rpath Add $LIBPATH to unit test eval | 29 July 2016, 09:13:10 UTC |
3c3862a | Zhou Wang | 19 July 2016, 07:34:31 UTC | Install Boost at /usr/local/boost-x.y.z on Docker Build unit tests only when Boost is available Configure Boost path in configure Address CR comments fix #678 This is a combination of 10 commits. Adapt makefile to work with Docker container on Linux This is a combination of 7 commits. enable unit tests build remove -liomp5 from math unit test enable openmp add -ldl fix space error add -dl for reader test change order of -l remove gdk installation use /usr/local/lib as boost library path Adapt path in unit tests on Linux adapt path for Linux adapt path on other test projects for Linux remove extra blank line add BOOST_PATH,and build unit tests based on BOOST_PATH configure boost path install Boost 1.60.0; use version variable instead of hardcoding. add comments for different paths on Linux than on Windows. restore installation of gdk, because the removal of gdk needs more changes and will be done in a separate check-in use ifdef, fix typos | 29 July 2016, 09:13:10 UTC |
180f593 | Zhou Wang | 18 July 2016, 14:46:45 UTC | Add Boost to Dockerfiles | 29 July 2016, 09:13:10 UTC |
39c60b5 | thilow | 28 July 2016, 21:20:38 UTC | ReduceLogSum: adapted core.bs. Tests still failing | 28 July 2016, 21:20:38 UTC |
b011011 | Frank Seide | 28 July 2016, 18:02:25 UTC | empty strings are now allowed in config CNTK variables | 28 July 2016, 18:02:25 UTC |
7397854 | Thilo Will | 28 July 2016, 15:49:52 UTC | ReduceLogSum backward path and core.bs | 28 July 2016, 15:49:52 UTC |
08691e9 | Mark Hillebrand | 28 July 2016, 15:27:35 UTC | Examples/SequenceToSequence/CMUDict/Data/cmudict-0.7b.mapping: normalize LE | 28 July 2016, 15:27:35 UTC |
38e4b2b | Thilo Will | 28 July 2016, 12:23:59 UTC | merged from master | 28 July 2016, 12:23:59 UTC |
540cd0b | Frank Seide | 27 July 2016, 22:11:57 UTC | addressed CR feedback | 27 July 2016, 22:11:57 UTC |
dde483f | thilow | 27 July 2016, 20:07:26 UTC | Adding ReduceLogSum | 27 July 2016, 20:07:26 UTC |
6a6ebc4 | U-FAREAST\fseide | 27 July 2016, 17:36:18 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into fseide/clonebs | 27 July 2016, 17:36:18 UTC |
da1f9f4 | Frank Seide | 27 July 2016, 17:35:17 UTC | fix in Convolution mapCount change | 27 July 2016, 17:35:17 UTC |
24cb169 | Mark Hillebrand | 27 July 2016, 11:31:50 UTC | Makefile: cppevalclient: link to stub libnvidia-ml.so (if real one not present) Fixes build of Tools/docker/Dockerfile-GPU (with regular docker). | 27 July 2016, 12:05:02 UTC |
8b68af3 | Frank Seide | 27 July 2016, 01:14:03 UTC | mapDims parameter to Convolution() is now optional, will be inferred from weight matrix | 27 July 2016, 01:14:03 UTC |
ee23bb2 | Frank Seide | 27 July 2016, 01:03:52 UTC | (fix for previous fix) | 27 July 2016, 01:03:52 UTC |
8f71698 | Frank Seide | 27 July 2016, 00:48:37 UTC | renamed reduction kernels that expect a hard-coded number of threads to reflect that number in their names | 27 July 2016, 00:48:37 UTC |
9dbf806 | Frank Seide | 26 July 2016, 23:51:19 UTC | merged from master | 26 July 2016, 23:51:19 UTC |
1a80a6a | Frank Seide | 26 July 2016, 20:55:12 UTC | undid accidental change of Shuffle() | 26 July 2016, 20:55:12 UTC |
5e357fe | Frank Seide | 26 July 2016, 20:52:21 UTC | addressed minor feedback from Amit's CR; addressed feedback from Simon Layton (NVidia) that the constants defined in GridDim are too small. | 26 July 2016, 20:52:21 UTC |
2c1b102 | Thilo Will | 26 July 2016, 11:59:58 UTC | fix 1BitDGD | 26 July 2016, 11:59:58 UTC |
6a6cff0 | Project Philly | 26 July 2016, 09:43:29 UTC | Integrate thilow/ReduceMax2 into master | 26 July 2016, 09:43:29 UTC |
4e4598e | Thilo Will | 26 July 2016, 08:27:37 UTC | merge with master | 26 July 2016, 08:27:37 UTC |
dcc7e9b | Thilo Will | 26 July 2016, 07:40:12 UTC | Added comments | 26 July 2016, 07:40:12 UTC |
759fa18 | Amit Agarwal | 26 July 2016, 05:43:11 UTC | CNTK v2 library: Add member method to NDMask to enable retreiving the buffer underlying a Mask object | 26 July 2016, 05:43:11 UTC |
3e6926d | Amit Agarwal | 25 July 2016, 23:33:25 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into amitaga/cntkv2LibraryOperators | 25 July 2016, 23:52:15 UTC |
575bbf6 | Amit Agarwal | 25 July 2016, 18:19:14 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into amitaga/cntkv2Library | 25 July 2016, 18:19:14 UTC |
524c527 | Thilo Will | 25 July 2016, 15:56:15 UTC | Using double as aggregator, hoping to fix issue with tests TWRGE TWRGS, TLRGS | 25 July 2016, 15:56:15 UTC |
52dda16 | Thilo Will | 25 July 2016, 14:06:03 UTC | Removed 'typedef' in partial specialisation of TensorOpReduction in hope to fix Linux build. | 25 July 2016, 14:06:03 UTC |
24c0ad1 | Thilo Will | 25 July 2016, 13:35:14 UTC | Fixed comments | 25 July 2016, 13:35:14 UTC |
2115db6 | Thilo Will | 25 July 2016, 13:17:13 UTC | Renamed variable to reductionOp | 25 July 2016, 13:17:13 UTC |
5dbb725 | Thilo Will | 25 July 2016, 13:00:11 UTC | factored aggregation op into a lambda | 25 July 2016, 13:00:11 UTC |
44d127e | Mark A. Hillebrand | 25 July 2016, 12:24:42 UTC | Normalize line endings where missing | 25 July 2016, 12:37:41 UTC |
4164856 | Amit Agarwal | 24 July 2016, 20:02:55 UTC | CNTK v2 library: Use dynamic loading of CompositeDataReader instead of a static dependency | 25 July 2016, 01:50:10 UTC |
a632393 | Amit Agarwal | 22 July 2016, 20:46:19 UTC | CNTK v2 library: Expose 18 additional operators in the v2 API | 23 July 2016, 01:41:48 UTC |
5b1c217 | Amit Agarwal | 22 July 2016, 23:46:09 UTC | CNTK v2 library: Added initial Trainer and MinibatchSource implementations, enabling end-to-end model training with v2 API. A test added to train a feed-forward network on the MNIST dataset using the v2 C++ API | 22 July 2016, 23:46:09 UTC |
a5d15f3 | U-FAREAST\fseide | 22 July 2016, 17:04:08 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into fseide/clonebs | 22 July 2016, 17:04:08 UTC |
7416f9b | Frank Seide | 22 July 2016, 16:54:13 UTC | bug fix: BatchNormalization::BackpropTo() should distinguish between saveMean and runMean the same way ForwardProp() does | 22 July 2016, 16:54:13 UTC |
27ff6f7 | Frank Seide | 22 July 2016, 15:53:31 UTC | (typo) | 22 July 2016, 15:53:31 UTC |
07a6fa2 | Frank Seide | 22 July 2016, 15:46:02 UTC | BatchNorm: moved allocation of saveMean() to where they are produced, and allocating them empty when they are not produced at all | 22 July 2016, 15:46:02 UTC |
0270010 | Frank Seide | 22 July 2016, 15:24:56 UTC | added new interface IFreezable to tell a node to freeze itself, in order to allow BatchNormalization to honor CloneFunction (..., parameters="constant") | 22 July 2016, 15:24:56 UTC |
fd95477 | Thilo Will | 22 July 2016, 15:02:33 UTC | Converted AggregationOp to pure function template | 22 July 2016, 15:02:33 UTC |
bd776dc | Thilo Will | 22 July 2016, 14:49:54 UTC | Changed "NeutralValue" to function templates | 22 July 2016, 14:49:54 UTC |
80fdb8f | Thilo Will | 22 July 2016, 14:28:53 UTC | using function overloading for neutral | 22 July 2016, 14:28:53 UTC |
fec05bf | Thilo Will | 22 July 2016, 13:51:15 UTC | Improved formatting and comments | 22 July 2016, 13:51:15 UTC |
ce350dd | Frank Seide | 22 July 2016, 02:47:45 UTC | (trying around with saveMean) | 22 July 2016, 02:47:45 UTC |
246a097 | Frank Seide | 22 July 2016, 02:39:10 UTC | (fixed an error message) | 22 July 2016, 02:39:10 UTC |
dcbb760 | Frank Seide | 22 July 2016, 02:11:02 UTC | bug fix: CloneFunction() should map output nodes, too | 22 July 2016, 02:11:02 UTC |