Граф коммитов

1122 Коммитов

Автор SHA1 Сообщение Дата
Friedel van Megen 8c90baf161 fix one more dependency 2017-03-20 11:06:48 +01:00
Chris Basoglu f36f278959 Update README.md 2017-03-18 18:04:25 -07:00
Emad Barsoum 8da12bcd49 Merge with master. 2017-03-17 10:36:40 -07:00
Alexey Reznichenko 124f07a0cc Another attempt at streamlining device selection
Update device selection API as agreed:
  * drop BestDevice() and DefaultDevice()
  * add an option to specify excluded devices that should not be used as
    globally default (neither auto-selected nor explicitly specified)
  * add a mechanism to query device properties
  * add a method to determine if a device is locked or not (in real-time)
  * in SetDefaultDevice, offer an option to exclusively lock the specified
    device.
2017-03-17 12:34:54 +01:00
Emad Barsoum f7f7eaf757 Fix one_hot in LU. 2017-03-16 22:58:13 -07:00
Emad Barsoum 2dc7df0922 More fixes from Jenkins. 2017-03-16 21:57:08 -07:00
Emad Barsoum 578504a0ad More fixes. 2017-03-16 19:33:12 -07:00
Emad Barsoum 72b9858189 Fix distributed training. 2017-03-16 18:31:09 -07:00
Emad Barsoum 8b47381d2f More fixes. 2017-03-16 17:25:17 -07:00
Emad Barsoum f341959d9d More ont_hot fixes. 2017-03-16 16:30:04 -07:00
Emad Barsoum 8af0ea3f18 Merge branch 'master' into 2.0rc/breakingChanges
Merge with master...
2017-03-16 14:19:31 -07:00
Zhou Wang 927c1e49cf add CreateSequence from sparse input
expose CreateSequence using sparse in C#

adapt API and C# binding

update baseline

aligning argument names

adapt test case to make sure at 1 non-zero values is created

adapt random generation of sparse sequence
2017-03-16 07:30:57 +01:00
Emad Barsoum e25dd44c70 Merge with master. 2017-03-15 22:10:17 -07:00
Emad Barsoum d86c231d1f Move all examples and notebook to use the new namespace. 2017-03-15 22:07:51 -07:00
Mark Hillebrand 51d02a4be5 Bump NuGet package versions 2017-03-15 17:17:07 +01:00
Mark Hillebrand 97e674c934 Bump versions 2017-03-15 17:16:54 +01:00
Otto Lu 0d6d05a330 Add node names for the `conv_bn_relu_layer` 2017-03-15 11:08:31 +08:00
Otto Lu 0c5f8052fb Update python script of bn-inception 2017-03-15 11:08:31 +08:00
Otto Lu 48e8d88810 Update `BN-Inception.cntk` 2017-03-15 11:08:31 +08:00
Otto Lu cd20b368cd update samples.json 2017-03-15 11:08:31 +08:00
Otto Lu 72223b0914 update model save path 2017-03-15 11:08:30 +08:00
Otto Lu a547656afe Update README.md 2017-03-15 11:08:30 +08:00
Otto Lu 3aa9492de0 add python version README.md 2017-03-15 11:08:30 +08:00
Otto Lu da396f8042 Update Python script of BN-Inception 2017-03-15 11:08:30 +08:00
Otto Lu 90025591f7 update brainscript of BN-Inception 2017-03-15 11:08:30 +08:00
Otto Lu 393b5672a1 Checkin Python example of BN-Inception 2017-03-15 11:08:30 +08:00
Cha Zhang 59671d547c Make convolution transpose work and code clean up. Also updated the Layers library.
Make convolution transpose work and code clean up. Also updated the Layers library.

Fixes for test.

Bug fix for test.

Changes based on CR.
2017-03-13 14:37:58 -07:00
David Pisani 1322f11a4e ResNet example changes: Save to model_path and restart switch added 2017-03-12 21:18:31 -07:00
Chris Basoglu caacc0b993 Fix documentation 2017-03-11 19:21:01 -08:00
Zhou Wang 7b8b9c1b74 add LoadModel from byte array
use istream

add tests

add check legacy model using memory buffer

make swig map char* type correctly

update test baseline

address CR comments
2017-03-11 21:34:12 +01:00
Junjie Qian bd93d7ec5a Copy gradients into a continous buffer in v2 aggregator 2017-03-10 09:15:40 -08:00
Ivan Rodriguez ab1bba45bc Adding Seq2Seq and HTKTruncated distributed tests. Also, adding 1bitsgd and block momentum tests for Alexnet example. 2017-03-09 10:30:57 +01:00
Project Philly 4abe7b11b9 Integrate zhouwang/cseval-example-clone into master 2017-03-07 04:00:01 -08:00
Amit Agarwal 730e592f39 CNTK v2 library: Improve error messages 2017-03-06 10:09:44 -08:00
Chris Basoglu 9ed8a4c725 Merge pull request #1498 from SaqibS/master
Fixing typos
2017-03-05 21:55:51 -08:00
Frank Seide 83f51bb34d another namespace fix 2017-03-03 19:02:20 -08:00
Frank Seide 260a7b076e added two missing Examples to PythonExamples project 2017-03-03 17:27:28 -08:00
Frank Seide 7c9fa5ac5c another example update 2017-03-03 17:27:10 -08:00
Frank Seide 1dcc6d3edc 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()
2017-03-03 17:21:54 -08:00
Frank Seide a59181fa76 setting up Trainer again without model=None as that may trigger a mem-sharing bug 2017-03-03 14:17:54 -08:00
Frank Seide b8bc888f43 removed one more Placeholder() from an example 2017-03-03 10:46:52 -08:00
Frank Seide a460a93190 reverted train/test_minibatch() to their old signature and behavior 2017-03-03 10:45:32 -08:00
Frank Seide 51a3ee5acb merged from master 2017-03-03 10:30:11 -08:00
Zhou Wang f969310afe add parallel evaluations to samples 2017-03-03 18:56:25 +01:00
Zhou Wang c7a5d3bc92 add examples for parallel evaluation 2017-03-03 18:30:00 +01:00
Zhou Wang 0424021d4a using share parameters for clone 2017-03-03 18:01:17 +01:00
Frank Seide dcdbe2005a fixed an import path after addressing CR feedback 2017-03-03 08:28:04 -08:00
Frank Seide 9bc2042885 updated some more tests after merge 2017-03-03 03:02:40 -08:00
Amit Agarwal 0514a40c55 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. 2017-03-02 22:32:07 -08:00
Frank Seide 20bfb76ed5 moved Evaluator from trainer.py to examples that use it, subject to a proper implementation 2017-03-02 20:51:09 -08:00