https://github.com/Microsoft/CNTK

sort by:
Revision Author Date Message Commit Date
942a6a4 gather gpu impl and ut 15 March 2017, 20:30:54 UTC
037d253 add gather op 09 March 2017, 21:53:56 UTC
22b836a move one hot to evaluation nodes 08 March 2017, 22:29:41 UTC
330d085 fix merge issue and add comments 08 March 2017, 19:46:11 UTC
376d92b update sparse and GPU sparse implementation 08 March 2017, 18:25:23 UTC
e08c135 add gpu dense matrix support, and UT 08 March 2017, 18:25:23 UTC
35b7edc add sparse support 08 March 2017, 18:25:22 UTC
4eb38c4 add one hot test op 08 March 2017, 18:25:21 UTC
be280b0 Integrate alrezni/learner_checkpoint into master 08 March 2017, 17:18:39 UTC
b024632 Integrate qiwye/asgd-fix into master 08 March 2017, 14:52:02 UTC
af44b12 Change learner's serialization/deserialization Get rid of the dependency on UIDs. Read/write smooth gradient values in the order of corresponding learner's parameters. Store them as a vector in checkpoint (as opposed to a dict from uids to values). 08 March 2017, 14:25:14 UTC
0e1b703 call install batch file instead of powerscript 08 March 2017, 11:31:03 UTC
6902ace fix install.ps1 called in ps subdirectory 08 March 2017, 09:24:00 UTC
d822b0f fix build: removing hard tabs 08 March 2017, 04:56:35 UTC
0df54d8 Enable python setup in debug bits 08 March 2017, 01:38:37 UTC
aafc06e Optimize the performane of CosineDistanceNode 08 March 2017, 01:38:37 UTC
383f3f3 start install powerscript through batch file 07 March 2017, 12:15:17 UTC
4abe7b1 Integrate zhouwang/cseval-example-clone into master 07 March 2017, 12:00:01 UTC
85bb5f0 Integrate kedeng/fixTimesReduceSeq into master 07 March 2017, 04:36:13 UTC
59009c2 Integrate vistepan/tensorboard-brainscript-support into master 07 March 2017, 03:30:04 UTC
37c7746 Fix dynamic axis in Times output when reducing sequence axis, and update test 07 March 2017, 03:08:03 UTC
b0453e6 Support for logging to TensorBoard from BrainScript. 07 March 2017, 01:38:34 UTC
020ed1a Fix rank-0 matrix shape to have dynamic axes in columns Also add a testcase for the cudaMalloc failure investigation 07 March 2017, 01:34:25 UTC
dda7e48 Integrate b023f6c85347c0e411abcd7123c15be4fb238ce0 into master 07 March 2017, 01:15:06 UTC
56b05eb Update README.md 06 March 2017, 23:06:21 UTC
8368586 Minor refactor and README update in CTC training 06 March 2017, 20:33:08 UTC
98eaedf Integrate amtiaga/bugFixesForReasonNet into master 06 March 2017, 19:09:54 UTC
730e592 CNTK v2 library: Improve error messages 06 March 2017, 18:09:44 UTC
b2dcdda Integrate eldak/sameStreamErrorMessage into master 06 March 2017, 18:02:35 UTC
47d3ffc CNTK v2 library: Minor bugfix to handle the situation of a sequence.reduce operation being performed on a placeholder operand whose shape is known but dynamic axes are unknown 06 March 2017, 17:53:38 UTC
d67957d Fixing error message when two streams with the same name are present 06 March 2017, 16:35:15 UTC
00b0b0c Merge branch 'master' into asgd-fix 06 March 2017, 15:40:44 UTC
b6e0fbf bug fix: CPU memory leak. 06 March 2017, 15:04:18 UTC
c735636 Making zip_open read only and optimization of conc_stack 06 March 2017, 11:37:40 UTC
1dc42fe Bug fix: change ASGD option from syncPeriod to SyncPeriodPerWorker 06 March 2017, 07:16:40 UTC
9ed8a4c Merge pull request #1498 from SaqibS/master Fixing typos 06 March 2017, 05:55:51 UTC
b023f6c Merge branch 'master' into pp_custom_metric 06 March 2017, 03:52:26 UTC
ba64f2c Merge branch 'master' into pp_custom_metric 06 March 2017, 03:50:55 UTC
ccca010 Update baseline for CTC MLF reader E2E tests, some refactoring 04 March 2017, 05:32:53 UTC
402b597 Add E2E test for MLF CTC reader 04 March 2017, 05:32:53 UTC
3e15371 Make phone boundaries in HTKMLFDeserializer configurable 04 March 2017, 05:32:53 UTC
0d1cf21 Extend MLF reader to read phone boundaries 04 March 2017, 05:32:53 UTC
83f51bb another namespace fix 04 March 2017, 03:02:20 UTC
260a7b0 added two missing Examples to PythonExamples project 04 March 2017, 01:27:28 UTC
7c9fa5a another example update 04 March 2017, 01:27:10 UTC
1dcc6d3 added a docstring example to past_value for data-dependent initial_state; one more round of deleting stale code and comments; fixed the slice bug of test_sequence_to_sequence() 04 March 2017, 01:21:54 UTC
c2cbc3d s2s test now sets the device before importing the example code, to avoid a device default mix-up 04 March 2017, 00:26:05 UTC
f90428c once again 03 March 2017, 23:34:22 UTC
b118973 another try at ThrowFormatted on gcc 03 March 2017, 23:33:25 UTC
07ba1d3 more trial and error 03 March 2017, 23:25:26 UTC
5768798 more gcc happiness 03 March 2017, 23:15:00 UTC
1257c61 (added an annotation) 03 March 2017, 23:06:46 UTC
aa0bc71 gcc happiness 03 March 2017, 23:04:55 UTC
a89cac3 bug fix: Linux version of ThrowFormatted() should not call itself 03 March 2017, 22:45:27 UTC
4407904 addressed CR feedback (wrong test in __getitem__()) 03 March 2017, 22:25:21 UTC
a59181f setting up Trainer again without model=None as that may trigger a mem-sharing bug 03 March 2017, 22:17:54 UTC
1bdd3da reverted depth_first_search distance/depth in stack variable to prior; deleted some unused left-over code; fixed some import/namespace issues with new tests 03 March 2017, 22:02:18 UTC
42256b6 updated 204 tutorial to new signature of LSTM() 03 March 2017, 19:50:23 UTC
a7553ba reverted and updated an accidentally committed over-the-top .gitignore 03 March 2017, 19:49:55 UTC
f5b86c2 increased tolerance of s2s test, due to random init 03 March 2017, 18:56:50 UTC
b8bc888 removed one more Placeholder() from an example 03 March 2017, 18:46:52 UTC
a460a93 reverted train/test_minibatch() to their old signature and behavior 03 March 2017, 18:45:32 UTC
51a3ee5 merged from master 03 March 2017, 18:30:11 UTC
f969310 add parallel evaluations to samples 03 March 2017, 17:56:25 UTC
a142bc2 add CPUONLY to the test project; simplify project configurations 03 March 2017, 17:31:54 UTC
c7a5d3b add examples for parallel evaluation 03 March 2017, 17:30:00 UTC
0424021 using share parameters for clone 03 March 2017, 17:01:17 UTC
2dcfab9 fixed a syntax error in a tutorial (merge error??) 03 March 2017, 16:44:46 UTC
7cfb0f8 fix after merge (missing import) 03 March 2017, 16:39:07 UTC
b685f7c removed debug messages 03 March 2017, 16:31:11 UTC
dcdbe20 fixed an import path after addressing CR feedback 03 March 2017, 16:28:04 UTC
435c880 Integrate kedeng/timesReduceSeq2 into master 03 March 2017, 13:23:39 UTC
4f645ba Integrate amtiaga/bugFixesForReasonNet into master 03 March 2017, 12:57:46 UTC
38615e2 Integrate v-yuxgu/ceil-pooling into master 03 March 2017, 12:52:29 UTC
9bc2042 updated some more tests after merge 03 March 2017, 11:02:40 UTC
bfd1197 bug fix: Python 2.7 overwrites iteration variable in list comprehension when it exists outside 03 March 2017, 10:45:49 UTC
755708b plenty of log messages for graph search; put that temporarily disabled test back in 03 March 2017, 09:57:29 UTC
7029d00 Times optimization for two sequences to avoid unroll in time steps This made attention sum calculation on CPU ~70x faster (from 180ms -> 2.5ms) in a case of attention sum for vocab_dim=40000, sequence_length=100, with 10 sequences in a batch 03 March 2017, 09:13:07 UTC
811208f 2.7 happiness in __call__() 03 March 2017, 08:33:25 UTC
054a00a changed to use uid for depth_first_search() visited set 03 March 2017, 08:23:31 UTC
0514a40 CNTK core: Made actual forward prop evaluation order consistent with the order in which we setup matrix sharing structure, to ensure proper evaluation of a set of nodes in accordance with the assumed evaluation order during allocating node value/gradient matrices. 03 March 2017, 06:32:07 UTC
578b535 temporarily removed test_sequence_grad_as_numpy_false, to see whether it is the test that makes the system hang 03 March 2017, 06:12:54 UTC
e8aba41 addressed CR feedback (renamed rf_shape [Cha]) 03 March 2017, 06:08:47 UTC
3e6f509 Merge branch 'master' into pp_custom_metric 03 March 2017, 05:45:05 UTC
3337b47 (minor fix of __getitem__()) 03 March 2017, 05:40:33 UTC
a24552b merged __getitem__() implementations in tensor.py 03 March 2017, 04:56:48 UTC
02e93d7 merged the two __getitem__() versions in tensor.py; added test for slice operator 03 March 2017, 04:55:26 UTC
20bfb76 moved Evaluator from trainer.py to examples that use it, subject to a proper implementation 03 March 2017, 04:51:09 UTC
77154a7 Integrate vistepan/progress-tracking-fixes into master 03 March 2017, 04:03:39 UTC
610065c adapted train_models_for_evaluation.py to run on CPU 03 March 2017, 03:55:57 UTC
be144fa update test modelversion 03 March 2017, 03:52:59 UTC
2d62f18 moved Variable.__getitem__() to TensorMixIn in prep for merge 03 March 2017, 03:26:11 UTC
f80fff2 Merge branch 'master' into v-yuxgu/ceil-pooling 03 March 2017, 02:13:09 UTC
410b090 Address feedbacks. 03 March 2017, 02:10:26 UTC
6256c6e Updated examples to use the latest progress writer API 03 March 2017, 01:57:57 UTC
486627d Merge branch 'master' into ebarsoum/prim1 For CR and checkin. 03 March 2017, 01:54:30 UTC
19f010c Shouldn't use std::swap in cuda. 03 March 2017, 01:53:30 UTC
6a4c9d6 reverted __call__() to old behavior 03 March 2017, 01:45:31 UTC
b141fc0 added some debug code for tracking down Python 2.7 __getattr__() problem; fixed accidental commit of build.bat 03 March 2017, 01:34:23 UTC
60cc5e2 updated a test 03 March 2017, 00:55:16 UTC
back to top