1aefd22 | Mark Hillebrand | 22 January 2016, 14:46:46 UTC | Change default targets for build-and-test | 22 January 2016, 14:46:46 UTC |
a0a3246 | Mark Hillebrand | 22 January 2016, 09:15:34 UTC | White-space | 22 January 2016, 09:15:34 UTC |
946bdc7 | Wolfgang Manousek | 22 January 2016, 09:11:05 UTC | Merge pull request #4 from Microsoft/mahilleb/Vssettings Update CppCntk.vssettings (wolfma) | 22 January 2016, 09:11:05 UTC |
5c0ccb6 | Mark Hillebrand | 22 January 2016, 09:08:52 UTC | Update CppCntk.vssettings (wolfma) | 22 January 2016, 09:08:52 UTC |
6f59eb1 | Mark Hillebrand | 22 January 2016, 08:23:03 UTC | Formatting | 22 January 2016, 08:23:03 UTC |
50c0e8e | Amit Agarwal | 22 January 2016, 06:00:54 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into amitaga/memoryFootprint | 22 January 2016, 06:00:54 UTC |
59287d0 | Dong Yu | 22 January 2016, 02:43:35 UTC | fix bug in imageLayout parameter and defaultHiddenActivity parameter in NDL usage prompt | 22 January 2016, 02:43:35 UTC |
8cf0b86 | Amit Agarwal | 22 January 2016, 02:33:40 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into amitaga/memoryFootprint | 22 January 2016, 02:33:40 UTC |
890dda9 | Dong Yu | 22 January 2016, 02:26:30 UTC | include imageLayout optional parameter in the usage prompt of Convolution, maxpooling, and avgPooling. in NDL | 22 January 2016, 02:28:44 UTC |
083dbac | Amit Agarwal | 22 January 2016, 02:14:05 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into amitaga/memoryFootprint | 22 January 2016, 02:14:05 UTC |
6137d50 | Amit Agarwal | 22 January 2016, 02:05:39 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into amitaga/preComputeProgressTracing | 22 January 2016, 02:05:39 UTC |
fd90701 | Frank Seide | 22 January 2016, 00:29:44 UTC | make gcc happy | 22 January 2016, 00:29:44 UTC |
c177d3a | Amit Agarwal | 22 January 2016, 00:26:51 UTC | Added progress tracing for PreCompute phase - currently just printing 0.0 instead of actual timeout which is sufficient for the current purpose | 22 January 2016, 00:26:51 UTC |
834e4e9 | U-FAREAST\fseide | 22 January 2016, 00:15:11 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into fseide/tensors | 22 January 2016, 00:15:11 UTC |
04f3089 | Frank Seide | 22 January 2016, 00:14:54 UTC | removed a now invalid consistency check | 22 January 2016, 00:14:54 UTC |
416ead4 | Frank Seide | 22 January 2016, 00:08:09 UTC | fixed verification of size of loaded learned parameter values: now works with non-matrices, too | 22 January 2016, 00:08:09 UTC |
08e4b99 | Alexey Kamenev | 21 January 2016, 23:19:48 UTC | Updated ResNet samples. | 21 January 2016, 23:19:48 UTC |
14fee02 | Frank Seide | 21 January 2016, 23:00:12 UTC | modified ComputationNode's matrix storage back to match Matrix dimensions if no minibatch, to ensure compatibility with sparse matrices | 21 January 2016, 23:00:12 UTC |
c9530e8 | Frank Seide | 21 January 2016, 20:59:47 UTC | updated PerformSVDecomposition() to use ValueAsMatrix() | 21 January 2016, 20:59:47 UTC |
ce0c087 | Frank Seide | 21 January 2016, 20:52:30 UTC | some consistency-check fixes related to SetDims() vs. MBLayout | 21 January 2016, 20:52:30 UTC |
8e7b19b | Frank Seide | 21 January 2016, 19:57:03 UTC | bug fix: LazyZeroGradient() must not use m_value dimensions for resizing m_gradient, because m_value may have been lost already due to memshare; bug fix: DelayedValueBase should not call SetDims(., false) because upon re-load (rollback) there is already an MB Layout, which will cause a consistency check to fail | 21 January 2016, 19:57:03 UTC |
2c9f56a | Amit Agarwal | 21 January 2016, 19:19:53 UTC | Added ability to ignore cudaFree return code in the tracing allocator since free is sometimes called during process exit when cudaFree calls would fail. | 21 January 2016, 19:19:53 UTC |
4691595 | Frank Seide | 21 January 2016, 18:31:55 UTC | bug fix, ClassBasedCrossEntropyWithSoftmax now working | 21 January 2016, 18:31:55 UTC |
797a9e0 | Amit | 21 January 2016, 02:01:08 UTC | Fixed linux build | 21 January 2016, 16:21:05 UTC |
5bbc37f | Frank Seide | 21 January 2016, 15:56:47 UTC | bug fix: TableLookupNode still used one wrong matrix interpretation | 21 January 2016, 15:56:47 UTC |
e389459 | Frank Seide | 21 January 2016, 07:28:33 UTC | LearnableParameter::InitRandom() now mimics the Matrix initialization behavior; bug fix: LookupTableNode used GetAsMatrixNumRows() on input[1] which is a minibatch; bug fix: Image/QuickE2E network definition updated to drop the now unnecessary extra column dimension; bug fix: TensorShape::IsDense() should not require m_offset to be 0 (column slices are perfectly fine) | 21 January 2016, 07:28:33 UTC |
d44c9ef | Frank Seide | 21 January 2016, 05:44:33 UTC | made gcc happy | 21 January 2016, 05:44:33 UTC |
f1a342c | U-FAREAST\fseide | 21 January 2016, 05:36:09 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into fseide/tensors | 21 January 2016, 05:36:09 UTC |
342b1ba | Frank Seide | 21 January 2016, 05:33:38 UTC | changed the mapping of tensors onto the Matrix storage objects: In case of no MBLayout, the Matrix objects now always have column dimension 1. Actual matrices (as used by TimesNode) are now 2D tensors. As part of this, GetNumCols()/-Rows() no longer exist, but were split into GetSampleMatrixCols()/-Rows() (minibatch interpretation) and GetAsMatrixCols()/-Rows() (2D matrix interpretation) | 21 January 2016, 05:33:38 UTC |
ffb930a | Amit Agarwal | 21 January 2016, 01:24:17 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into amitaga/memoryFootprint | 21 January 2016, 01:24:17 UTC |
2839083 | Amit Agarwal | 21 January 2016, 01:20:12 UTC | Added free memory reporting as part og GPU memory allocation/free traces and some refactoring | 21 January 2016, 01:20:12 UTC |
7712bea | Amit Agarwal | 20 January 2016, 20:15:51 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into amitaga/memoryFootprint | 20 January 2016, 20:15:51 UTC |
31a1646 | Alexey Kamenev | 20 January 2016, 19:04:31 UTC | Fixed topk memory leak. | 20 January 2016, 19:07:16 UTC |
c977f39 | Philipp Kranen | 20 January 2016, 09:42:17 UTC | Update README.md | 20 January 2016, 09:42:17 UTC |
98c5edd | Philipp Kranen | 20 January 2016, 09:34:03 UTC | Update README.md | 20 January 2016, 09:34:03 UTC |
9996e69 | Philipp Kranen | 20 January 2016, 09:23:52 UTC | Merge pull request #3 from Microsoft/mahilleb/readme-patch Update README.md | 20 January 2016, 09:23:52 UTC |
6967edd | Mark Hillebrand | 20 January 2016, 09:18:30 UTC | Update README.md | 20 January 2016, 09:18:30 UTC |
213294b | Philipp Kranen | 20 January 2016, 08:59:19 UTC | Update README.md | 20 January 2016, 08:59:19 UTC |
5b718d2 | Philipp Kranen | 20 January 2016, 08:43:21 UTC | Added perf chart for entry page | 20 January 2016, 08:43:21 UTC |
cc21bb7 | Amit Agarwal | 20 January 2016, 07:01:31 UTC | Fixed CPUONLY build | 20 January 2016, 07:29:28 UTC |
cf14913 | Amit Agarwal | 20 January 2016, 06:19:18 UTC | Added CUDA memory allocation/free tracing facility to GPUMatrx and GPUSparseMatrix | 20 January 2016, 06:19:18 UTC |
01bdefe | U-FAREAST\fseide | 19 January 2016, 17:08:08 UTC | Merge branch 'master' of https://github.com/Microsoft/CNTK into fseide/tensors | 19 January 2016, 17:08:08 UTC |
6781765 | Frank Seide | 19 January 2016, 16:49:26 UTC | merged fseide/tensors and master after github move | 19 January 2016, 16:49:26 UTC |
1ae3404 | Frank Seide | 19 January 2016, 00:50:09 UTC | moved ParallelNode to EsotericNodes.h (it is redundant with RowStackNode) | 19 January 2016, 00:50:09 UTC |
78a7ede | Frank Seide | 19 January 2016, 00:40:31 UTC | BestGpu::Init() no longer throws when run under remote access | 19 January 2016, 00:40:31 UTC |
742e239 | Amit Agarwal | 18 January 2016, 23:03:40 UTC | Minor: Pending CR feedback - fixed some typos in comments | 18 January 2016, 23:03:40 UTC |
ee1fe53 | Mark Hillebrand | 18 January 2016, 10:27:26 UTC | Tune README.md | 18 January 2016, 10:27:26 UTC |
4e1f857 | Mark Hillebrand | 18 January 2016, 10:25:45 UTC | Update README.md | 18 January 2016, 10:25:45 UTC |
65f4b48 | Mark Hillebrand | 18 January 2016, 10:23:34 UTC | .gitignore: ignore ReaderTests output | 18 January 2016, 10:23:34 UTC |
f1c3232 | Mark Hillebrand | 18 January 2016, 08:40:10 UTC | Missing tabs-to-spaces conversion | 18 January 2016, 08:40:10 UTC |
6c5aa58 | Mark Hillebrand | 18 January 2016, 08:36:24 UTC | Add submodule for 1bit-SGD This is available under a different license, see https://github.com/Microsoft/CNTK-1bit-SGD for more information. | 18 January 2016, 08:36:24 UTC |
dcfc0e9 | Mark Hillebrand | 18 January 2016, 08:36:17 UTC | License change | 18 January 2016, 08:36:17 UTC |
2f39fe3 | Mark Hillebrand | 18 January 2016, 08:36:14 UTC | Re-format code using clang-format (plus some post-processing) | 18 January 2016, 08:36:14 UTC |
1bcefe7 | Mark Hillebrand | 18 January 2016, 08:35:54 UTC | .gitattributes: set proper configuration for text files (and normalize files) | 18 January 2016, 08:35:54 UTC |
30b3f3b | Mark Hillebrand | 18 January 2016, 08:33:56 UTC | License change | 18 January 2016, 08:33:56 UTC |
146de26 | Amit Agarwal | 18 January 2016, 06:00:31 UTC | Merge branch 'master' of https://git01.codeplex.com/cntk into amitaga/memshare | 18 January 2016, 06:00:31 UTC |
4f2a597 | Amit Agarwal | 18 January 2016, 05:44:19 UTC | Bugfix: Enabled output value memory sharing for non-BS CNTK configs too | 18 January 2016, 05:44:19 UTC |
fa4be28 | U-FAREAST\fseide | 18 January 2016, 01:44:14 UTC | Merge branch 'master' of https://git.codeplex.com/cntk into fseide/tensors | 18 January 2016, 01:44:14 UTC |
97fadca | Frank Seide | 18 January 2016, 01:42:35 UTC | added some consistency checks for m_numCols in prep for removing it | 18 January 2016, 01:42:35 UTC |
3d87c90 | Amit Agarwal | 18 January 2016, 01:11:28 UTC | Added a temporary comment about about the MPI v7 requirement for MPI_Iallreduce call on Windows. This comment will be removed in a few weeks. | 18 January 2016, 01:11:28 UTC |
e37fede | Frank Seide | 18 January 2016, 00:20:01 UTC | updated the link in MPIWrapper.h to the required MPI SDK installer; ComputationNode::GetNumRows() now returns its value directly from m_sampleLayout, the m_numRows member has been deleted | 18 January 2016, 00:20:01 UTC |
582b03d | Amit Agarwal | 17 January 2016, 02:13:03 UTC | Some minor node value matrix memory sharing fixes | 17 January 2016, 02:13:03 UTC |
1805830 | Frank Seide | 16 January 2016, 08:07:38 UTC | made gcc happy (ptrdiff_t vs. int) | 16 January 2016, 08:07:38 UTC |
b933d83 | U-FAREAST\fseide | 16 January 2016, 07:53:06 UTC | Merge branch 'master' of https://git.codeplex.com/cntk into fseide/tensors | 16 January 2016, 07:53:06 UTC |
267be9e | Frank Seide | 16 January 2016, 07:50:06 UTC | completed ShiftNode::BackpropTo() | 16 January 2016, 07:50:06 UTC |
c186d49 | Frank Seide | 16 January 2016, 01:08:26 UTC | added specializations to linear unary non-reducing ops: copy and add ops now use existing CUDA functions; ShiftNode backprop functional for non-packed case/no boundary state | 16 January 2016, 01:08:26 UTC |
eb0e08e | Amit Agarwal | 15 January 2016, 22:31:32 UTC | Fixed testcases.yml to use build SKU gpu instead of default to align with the previous build SKU option value change in TestDriver | 15 January 2016, 22:31:32 UTC |
4ac49b0 | Frank Seide | 15 January 2016, 19:21:25 UTC | ShiftNode::ForwardProp() implemented | 15 January 2016, 19:21:25 UTC |
d5a3a89 | Amit | 15 January 2016, 18:38:59 UTC | Changed default build location setting for Linux to accomodate multiple build SKUs | 15 January 2016, 18:44:52 UTC |
8ae5176 | Amit Agarwal | 15 January 2016, 17:40:43 UTC | Merge branch 'master' of https://git01.codeplex.com/cntk into amitaga/separate1bitDataParallelSGD | 15 January 2016, 17:40:43 UTC |
a8029ea | Amit Agarwal | 15 January 2016, 17:40:00 UTC | Changed TestDriver.py --build-sku option values | 15 January 2016, 17:40:00 UTC |
9241f30 | Mark Hillebrand | 15 January 2016, 15:32:49 UTC | README.md: minor fixes | 15 January 2016, 15:32:49 UTC |
e9a91a8 | bmitra | 15 January 2016, 11:30:27 UTC | Merge branch 'master' of https://git01.codeplex.com/cntk into bmitra/Changes | 15 January 2016, 11:30:27 UTC |
3ce17cb | bmitra | 15 January 2016, 11:30:15 UTC | Bug fix for _tensorShuffleScaleAndAddRowSparse kernel. | 15 January 2016, 11:30:15 UTC |
58d3987 | Amit Agarwal | 15 January 2016, 09:10:29 UTC | Added missing parameter to matchesTag call in the run test command | 15 January 2016, 09:10:29 UTC |
e32329a | Amit Agarwal | 15 January 2016, 08:08:17 UTC | Merge branch 'master' of https://git01.codeplex.com/cntk into amitaga/separate1bitDataParallelSGD | 15 January 2016, 08:08:17 UTC |
252abe2 | Amit Agarwal | 15 January 2016, 08:08:07 UTC | Added --build-sku option to TestDriver.py to enable filtering tests based on build SKUs (aka CPUONLY vs GPU vs 1BitSGD) | 15 January 2016, 08:08:07 UTC |
e4adfb7 | Alexey Kamenev | 14 January 2016, 20:53:41 UTC | Updated readme.txt | 14 January 2016, 20:53:41 UTC |
9d26271 | Alexey Kamenev | 14 January 2016, 18:49:33 UTC | Updated ResNet samples. | 14 January 2016, 20:32:48 UTC |
0f2786c | Amit | 14 January 2016, 18:58:31 UTC | Added support for 1BitSGD build flavor to build-and-test script. Also fixed a CPUONLY build error | 14 January 2016, 18:58:31 UTC |
dd858fa | bmitra | 14 January 2016, 17:46:31 UTC | Minor GPU sparse changes. | 14 January 2016, 17:46:31 UTC |
a1dc007 | bmitra | 14 January 2016, 17:02:48 UTC | Adding IsValid check and minor spell corrections in unit tests. | 14 January 2016, 17:02:48 UTC |
b3f7361 | Jasha Droppo | 08 January 2016, 22:34:29 UTC | TrainOrAdaptModel, when doing regularized adaptation, now properly INvalidateCompiledNetwork() before calling CompileNetwork() again. | 14 January 2016, 17:00:04 UTC |
f85ef91 | Jasha Droppo | 08 January 2016, 22:22:42 UTC | FromEvalOrder() warning message had possible nullptr dereference, fixed. | 14 January 2016, 17:00:03 UTC |
ada6b57 | bmitra | 14 January 2016, 15:41:43 UTC | Minor sparse changes. | 14 January 2016, 15:41:43 UTC |
8d646e1 | Philipp Kranen | 14 January 2016, 10:12:20 UTC | removing zombie vcxproj files | 14 January 2016, 10:12:20 UTC |
135b992 | Philipp Kranen | 11 January 2016, 17:19:07 UTC | fixed Linux Makefile wrt deleted files | 14 January 2016, 08:06:40 UTC |
3eeb187 | Philipp Kranen | 11 January 2016, 13:18:13 UTC | reflected deleted files and example updates in VS solution | 14 January 2016, 08:06:40 UTC |
9e9cab0 | Philipp Kranen | 11 January 2016, 12:52:31 UTC | removing files and folders as announced (examples and readers) | 14 January 2016, 08:06:40 UTC |
fc61bb1 | Philipp Kranen | 18 December 2015, 14:28:51 UTC | Removed a few unreferencep source files | 14 January 2016, 08:03:10 UTC |
18528f1 | Amit Agarwal | 14 January 2016, 06:43:27 UTC | Merge branch 'master' of https://git01.codeplex.com/cntk into amitaga/separate1bitDataParallelSGD | 14 January 2016, 06:43:27 UTC |
40cbeac | Frank Seide | 14 January 2016, 03:32:49 UTC | more strict control of compiled/non-compiled state of networks in MEL. All editing operations invalidate compilation, and Saving the model will validate the model if changed (if not it was validated during Load) | 14 January 2016, 03:32:49 UTC |
c89b632 | U-FAREAST\fseide | 14 January 2016, 02:56:33 UTC | Merge branch 'master' of https://git.codeplex.com/cntk into fseide/tensors | 14 January 2016, 02:56:33 UTC |
b3cfd6e | Frank Seide | 14 January 2016, 02:54:01 UTC | bug fix in TensorSliceWithMBLayoutFor() handling of m_timeOffset for entire batch | 14 January 2016, 02:54:01 UTC |
31b0e70 | Frank Seide | 14 January 2016, 02:42:17 UTC | added a new file-format option to the MEL LoadModel command in order to load older CNTK files that had borked tensor dimensions; FrameRange now allows time ranges | 14 January 2016, 02:42:17 UTC |
de5be29 | thhoens | 13 January 2016, 00:23:04 UTC | Minor performance upgrade on row slicing to avoid GPU mem copy. | 13 January 2016, 23:45:20 UTC |
f835efd | thhoens | 13 January 2016, 00:01:29 UTC | Fix for multi GPU to share all parameters required to adjust learning rate. | 13 January 2016, 23:45:19 UTC |
d39d87f | thhoens | 09 January 2016, 12:24:23 UTC | Fixed a bug where the m_elemSizeAllocated was used instead of m_nz | 13 January 2016, 23:45:18 UTC |
b66ea69 | Amit Agarwal | 13 January 2016, 22:46:33 UTC | Remove 1bit SGD implementation headers references from the CNTK project files | 13 January 2016, 22:46:33 UTC |
9cf113d | Amit | 13 January 2016, 22:29:19 UTC | Made 1bit support an optionally configurable build option for CNTK which is off by default | 13 January 2016, 22:29:19 UTC |