253e65b432
ReduceLogSum: beautifications
2016-08-05 14:08:54 +02:00
7397854908
ReduceLogSum backward path and core.bs
2016-07-28 17:49:52 +02:00
c508e9f4ca
CNTK v2 library: Sparse input support
2016-06-11 02:12:39 -07:00
c140a1e264
Adding Compare nodes (Less, Greater, ...)
2016-05-23 16:06:40 +02:00
25eff37ba6
merged with master
2016-04-26 11:05:58 +02:00
7d1cfb77e6
merged with master
2016-04-25 17:41:54 +02:00
25eba71b53
addressing CR comments, implementing Nnary zip validation
2016-04-25 12:30:08 +02:00
a4867f2b5c
renaming gradient op to be more generic
2016-04-25 12:30:01 +02:00
87597779c7
addressing first set of comments from CR
2016-04-25 12:27:26 +02:00
7c44bbb2f2
implementation of clipbyvalue
2016-04-25 12:10:51 +02:00
a794383152
renaming gradient op to be more generic
2016-04-25 12:08:34 +02:00
04ba892adf
addressing first set of comments from CR
2016-04-25 12:07:21 +02:00
3efd2197af
implementation of clipbyvalue
2016-04-25 12:01:59 +02:00
309c8bd26e
Review changes
2016-04-22 08:44:56 -07:00
fac52a5cc4
Review changes
2016-04-22 08:43:41 -07:00
2d5614618d
Floor,ceil,round operators impl. and tests
2016-04-22 08:43:41 -07:00
18e8f49f80
Implementing IfNode
2016-04-22 10:17:13 +02:00
7e8008b818
addressed code-review minor feedback
2016-04-14 11:40:46 -07:00
199937d212
simplified DISPATCH_MATRIX_ON_FLAG_USEBOTH_4BOTH (code-review feedback)
2016-04-13 13:37:27 -07:00
cf101570ec
changes to support sin() node
2016-04-07 16:16:11 +02:00
157bf77be1
Cleanup of old code.
2016-04-06 11:03:48 -07:00
b3d024f9c5
Fixed nested template shadowing.
2016-04-05 16:09:26 -07:00
fb1eb58c02
Added comments and fixes related to Amit's comments.
2016-04-05 14:09:12 -07:00
c855efaa83
Fixed with Frank's addition to the CR.
2016-04-04 21:02:19 -07:00
9068877a07
Merged with master and fixed conflicts.
2016-04-04 19:14:31 -07:00
ec6c8595f3
Changed Matrix to use shared_ptr instead of raw pointers.
2016-04-04 13:52:50 -07:00
ca4380ecd8
Merge remote-tracking branch 'origin/master' into thhoens/matrixmerge
...
Conflicts:
Source/1BitSGD
Source/Math/CPUMatrix.cpp
Source/Math/CPUMatrix.h
Source/Math/CuDnnConvolutionEngine.cu
Source/Math/GPUMatrix.h
Source/Math/Matrix.cpp
Source/SGDLib/SimpleEvaluator.h
Tests/EndToEndTests/ModelExport/Model0/baseline.txt
Tests/EndToEndTests/ModelExport/Model0/testcases.yml
Tests/EndToEndTests/ModelExport/Model1/baseline.txt
Tests/EndToEndTests/ModelExport/Model1/testcases.yml
2016-04-01 22:39:56 -07:00
a285934411
Updated for Linux build errors.
2016-04-01 16:08:10 -07:00
3efbc25015
Changed SetBuffer to requrie the size, allocated. Fixed a bug in resize of GPUSparse.
2016-04-01 13:46:36 -07:00
c02829cf0f
Renamed BufferPointer to Data and GetArray to Buffer. Also began changing Resize to RequireSize to fix VerifyResizable.
2016-03-30 22:25:25 -07:00
a3bc1fdd8e
Replaced int* with Matrix<int> in ReferenceConvolutionEngine, updated unit tests.
2016-03-30 09:55:34 -07:00
69e3971426
Added feed forward CPU implementation for reference convo engine.
2016-03-30 09:55:06 -07:00
b72fbc61ba
refactoring towards allowing MB data for first argument of Times
2016-03-25 11:03:47 -07:00
e0aa22181b
Added the fixes from the Code Review.
2016-03-24 16:15:53 -07:00
59385cf3ea
align some more backslashes
2016-03-22 13:20:00 +01:00
fe447bff01
addressing reviewer comments (mainly formatting issues)
2016-03-22 13:19:59 +01:00
25025884b3
first attempt at logplusnode
2016-03-22 13:18:49 +01:00
f94cc3deea
Fix for Linux
2016-03-21 15:46:18 -07:00
b4a84115e4
Initial commit for Jenkins
2016-03-21 13:55:24 -07:00
eb520f4310
re-applying changes for reciprocal and elementdivide in BrainScript with new structure in NonlinearityNodes.h
2016-03-21 15:27:50 +01:00
2e03963209
added code for element-wise Sqrt() node
2016-03-17 08:48:34 +01:00
a6a80cded9
add support for node Abs; take CR comments into account
2016-03-09 11:57:56 +01:00
7cebf329bb
merged with master
2016-03-04 16:18:17 -08:00
b6a5bfe6f5
bug fix in var estimation: must sqr first and reduce after, and must launch with correct beta in reduction
2016-03-04 15:41:16 -08:00
2fad8dd410
Added comments from CR.
2016-03-04 00:28:54 -08:00
8199bd13f9
Removed matrix name and unused Read from file constructor as per discussion with fseide.
2016-03-04 00:28:52 -08:00
359d90ab09
full tensor support for Mean and InvStdDev operations
2016-03-03 23:46:17 -08:00
713b8bf952
gcc stack traces are now formatted by inserting double spaces around the actual function name, to make for easier visual parsing
2016-03-02 10:52:46 -08:00
673506e8ca
some clean-up of sparse matrices;
...
disabled meaningless elementwise sparse ops such as log();
sparse matrice slices now keep track of their original matrix (not used yet);
bug fix: ~SyncGuard() should not swallow errors since CUDA errors may get discovered here;
bug fix: HTKMLFReader test 16 must use consistent precision
2016-03-01 10:43:50 -08:00
e70b955230
Fixed hostname output in CudaCall and also made the CUDA device id reporting in CudaCall accurate by using cudaGetDevice
2016-01-30 11:10:57 -08:00
Thilo Will
Amit Agarwal
William Darling
Aayush Garg
Frank Seide
thhoens
Alexey Kamenev