ebf1ab9f8b
Add a function such that cn.exe will print build signature (build path, builder, build time, build branch) on start-up.
...
(useful when there are a lot of versions)
It assumed environment "%USERNAME% is set and "hostname" is in path, which
should be true on most machines.
If users have not put "git" in their PATH, branch and commit SHA1 will not be
shown.
2015-02-01 23:57:55 -08:00
986fac41ca
Using verbosity in HTKMLFReader to control the output.
...
When verbosity==0, no more "getting utterances ... " and "feature set * : paging in randomized chunk" output
2015-02-01 23:31:46 -08:00
b8dd80c67c
When nbruttsineachrecurrentite is larger than 1, learnratePerSample is normalized
2015-02-01 23:12:00 -08:00
1a6b330048
1) do not redirect stderr in debug mode ; 2) pre-computing node is cacluated on all the data instead of the first epoch (there is an option in config to fallback to the old behaviour: using only the first epoch data)
2015-02-01 23:07:41 -08:00
d424e6bcf7
Merge remote-tracking branch 'origin/master' into linux-gcc
2015-01-31 15:17:31 -05:00
53df82fdc4
SequenceParser with large buffer for reading strings.
2015-01-30 23:02:36 -08:00
e4345883d6
Have a large MAXSTRING
2015-01-30 16:15:13 -08:00
bba6a97f06
Merge branch 'master' of https://git01.codeplex.com/cntk
2015-01-30 15:38:29 -08:00
d7a6bc27d3
GetIdFromLabel should only used for retrieval. Return error if a label is not found in vocabulary.
2015-01-30 15:37:10 -08:00
28a8f4dc46
Add TimeUtility to the makefile.
2015-01-29 17:18:59 -05:00
59548f97f7
change UINT64 to unsigned long long (no uint64 in linux).
2015-01-29 17:15:17 -05:00
ad971e99c7
Merge remote-tracking branch 'origin/master' into linux-gcc
2015-01-29 16:07:38 -05:00
924d79a46e
Fix the compile issue when use "byte" in GPUSparseMatrix.cu
2015-01-29 16:06:49 -05:00
73e20eaac2
Allow named arguments to take non literal values in macros.
2015-01-29 10:22:07 -08:00
ac50417b75
fix the build problem in the debug build of CNTKEval project.
2015-01-28 23:21:59 -08:00
0cbab9b3e7
move TimerUtilities.h to common\include. Add TimerUtilities.cpp to CNTKEval project.
2015-01-28 15:59:31 -08:00
97a317e861
force gradient to be sparse if input is sparse in the TimesNode. This is the only dense X sparse combination we suppport right now.
2015-01-28 11:18:16 -08:00
5fd74fe9a3
fill constant node's value in second pass if the size of the constant is known by then.
2015-01-27 15:38:06 -08:00
7baca24d75
changed timing function.
2015-01-27 14:35:20 -08:00
fe68c14b9d
Add explicit template instantiation for TaskDescriptor and update Makefile for other readers.
2015-01-27 16:41:54 -05:00
8fbcf8f5b7
added functionality for assymmetric context windows
2015-01-27 09:53:42 -08:00
f9c555830b
added functionality for asymmetric context windows
2015-01-27 09:47:26 -08:00
13957ff777
fix -O2 problem
2015-01-27 15:31:10 +08:00
aecdb3503c
Merge branch 'linux-gcc' of https://git.codeplex.com/cntk into linux-gcc
2015-01-24 13:47:46 -05:00
508d1bf081
Fix compile error for g++ (it cannot automatic infer some types like icc).
2015-01-24 13:45:26 -05:00
0f42a7d349
A temporary fix for Linux and Windows wstring handls.
...
Details:
Change fputstring function, for linux, always convert a wstring to a char16_t and then write it.
Change fgetstring function, for linux, use fgetc twice to get a char16_t.
2015-01-24 13:42:42 -05:00
7ed0c8c396
Modified readme for ACML.
2015-01-24 13:39:40 -05:00
8c61da007a
Merge branch 'master' of https://git01.codeplex.com/cntk
2015-01-23 15:30:23 -08:00
337033f65d
Change FindSymbol to fix a MEL bug that occurs when using NDL command with dotted names.
2015-01-23 15:30:06 -08:00
cf91c565e3
Merge remote-tracking branch 'origin/master' into linux-gcc
...
update linux-gcc to newest version
2015-01-23 15:47:18 -05:00
d873d8b82d
Merge remote-tracking branch 'origin/master' into linux-gcc
...
Get the updated version for linux-gcc
2015-01-23 15:35:11 -05:00
471599e15f
More changes to Makefiles. Eventually, they probably could be merged.
2015-01-22 17:23:36 -05:00
8b8f20138b
Fix the bug during loadCheckPointFile. It try to putMarker in the old file.
2015-01-21 23:03:28 -05:00
032a0e8ede
Merge branch 'master' of https://git01.codeplex.com/cntk
2015-01-21 18:25:58 -08:00
6e1876111c
Change Sparse Matrix implementation (esp. GPU Sparse matrix) to make it work for LM training and other sparse features.
...
Note: still many improvements are needed for sparse matrices.
2015-01-21 18:25:41 -08:00
169d8b14bc
changes to Makefiles for cleaner builds.
2015-01-21 17:39:15 -05:00
09afb0dd9e
changes to Makefile.cpu, Makefile.gpu and CPUMatrix.cpp to have it compile in linux with ACML library.
2015-01-21 14:56:49 -05:00
de39c79e4f
Added README.
2015-01-20 14:41:29 +08:00
dc6164c06a
Merge branch 'master' of https://git01.codeplex.com/cntk
2015-01-19 16:30:33 -05:00
0879cfef56
Fix the bug for MEL when adding a recurrent loop.
2015-01-19 16:30:25 -05:00
08a9c5e946
Add plumbing for ExpNode in NDL
2015-01-19 09:27:41 -08:00
8aef179d7d
Creating new branch linux-gcc
...
Merge branch 'master' of https://git.codeplex.com/cntk into linux-gcc
2015-01-18 15:08:34 +08:00
43ea68b59e
change SparseInputValue node and CPU Sparse matrix to make LM CPU training work.
2015-01-17 13:37:00 -08:00
0443dc818d
Merge branch 'master' of https://git.codeplex.com/cntk
...
Conflicts:
.gitignore
Common/Include/fileutil.h
2015-01-17 21:40:37 +08:00
b5c41e6516
A large modification, allowing compilation of CNTK under linux g++
...
1. Add Common/Include/Platform.h for substitution of windows API and types
2. Fixed compile errors under g++, warning is not handled.
3. Compiles Readers into shared library, added code in *Reader/Export.cpp for dynamic linking and loading
4. Fixed msra::strfun::_strprintf to make it work under linux g++
5. Added makefiles for building (cpu and gpu)
2015-01-17 21:03:40 +08:00
459ff2ded1
Merge branch 'master' of https://git01.codeplex.com/cntk
2015-01-14 22:51:55 -05:00
97d5fb415a
FIx the bug when Dropout node in recurrent loop. Bug: the funcationValues is unintialized when dropOutNode in a recurrent loop.
2015-01-14 22:23:57 -05:00
eafc86daab
MPI Model Averaging. Added check to short-circuit empty minibatch
2015-01-14 18:51:05 -08:00
3e68d18ad8
Improvements to MPI Model Averaging.
...
DecimateMinibatch now checks that all input matrices have the same number of columns, and throws an exception if they do not.
DecimateMinibatch now checks for empty input matricies (after decimation).
The reduction/averaging code now weights each model by the relative number of samples in it's effective minibatch.
TrainOneEpoch now returns the local value totalEpochSamples, so the proper weighting can be performed during model updating.
2015-01-14 18:09:39 -08:00
9e37c72050
Changed Deviceid=auto to be MPI model-averaging compatable, no more device contention.
2015-01-14 18:09:38 -08:00
erw
Yu
kaisheny
Dong Yu
Hakan Erdogan
Mike Seltzer
wengong_jin
yzhang87
Jasha Droppo
ascii991218