- 23 6月, 2016 19 次提交
-
-
由 A. Unique TensorFlower 提交于
Change: 125610051
-
由 A. Unique TensorFlower 提交于
Change: 125608314
-
由 Gunhan Gulsoy 提交于
Change: 125604719
-
由 Eugene Brevdo 提交于
Change: 125604684
-
由 A. Unique TensorFlower 提交于
Change: 125601282
-
由 Vinu Rajashekhar 提交于
- Earlier we were locally copying the map which could be potentially very big. Change: 125595701
-
由 A. Unique TensorFlower 提交于
Factors out the interface for control flow ops Enter, Exit, Switch, Merge and NextIteration. Registers the Gather op for quantized parameters. Change: 125593594
-
由 Vijay Vasudevan 提交于
in a subdirectory under the workspace. Change: 125590691
-
由 A. Unique TensorFlower 提交于
Change: 125589121
-
由 A. Unique TensorFlower 提交于
Change: 125584342
-
由 A. Unique TensorFlower 提交于
Change: 125584181
-
由 A. Unique TensorFlower 提交于
Change: 125582830
-
由 Alexey Surkov 提交于
When the range can't be satisfied, the GCS server returns the 416 HTTP code and some additional information in the body. The curl library expects the exact amount of bytes in the body as requested in the range, so curl_easy_peform fails with a message like "Failed writing body (12 != 171)". With this fix, we don't check the return value of curl_easy_peform if the HTTP code was 416. Change: 125579779
-
由 Eugene Brevdo 提交于
Change: 125578894
-
由 A. Unique TensorFlower 提交于
Change: 125575345
-
由 Gunhan Gulsoy 提交于
and tensorflow/core/distributed_runtime:remote_device_test* from opensource build. Change: 125573873
-
由 A. Unique TensorFlower 提交于
local_init_op are run by each worker, while the init_op is only run by the chief. Fix __all__ Added initial README.md Change: 125573668
-
由 A. Unique TensorFlower 提交于
Change: 125572152
-
由 A. Unique TensorFlower 提交于
Change: 125571683
-
- 22 6月, 2016 21 次提交
-
-
由 A. Unique TensorFlower 提交于
Fixing bug in which the default value of the init_op was checked against None rather than _USE_DEFAULT. This wasn't caught as the value of _USE_DEFAULT matched supervisor._USE_DEFAULT. A unit test has been added which fails for the old code but succeeds with the new code. Change: 125564252
-
由 A. Unique TensorFlower 提交于
Use a function to compute dimensions for backpropagation, rather than using a macro. Return dimensions using a struct rather than individual ints. Change: 125556701
-
由 Manjunath Kudlur 提交于
Fixes #2974 Change: 125530598
-
由 A. Unique TensorFlower 提交于
Change: 125511710
-
由 Eugene Brevdo 提交于
Change: 125510452
-
由 A. Unique TensorFlower 提交于
Change: 125508441
-
由 A. Unique TensorFlower 提交于
Change: 125496972
-
由 A. Unique TensorFlower 提交于
Add partitioner that partitions a variable such that no partition gets less than the given minimum size of chunk. Swith from 'layers.legacy_fully_connected' to 'layers.fully_connected' in _DNNLinearCombinedBaseEstimator. Also, partition the weights of the fully connected layers using the added partitioner. Change: 125492475
-
由 Illia Polosukhin 提交于
* optimize_loss expects 0d Tensor. * Estimator.get_eval_ops requires to have `targets` not None. * Adding asserts for shapes in rnn_cells. * Added error check in rnn.rnn. Change: 125491759
-
由 Vijay Vasudevan 提交于
Change: 125491544
-
由 Cassandra Xia 提交于
Added tf-storage component for storing TensorBoard URI state and modified tf-tensorboard.html to use it. Added tf-globals component to hold TensorBoard global variables. Change: 125489418
-
由 Adria Puigdomenech 提交于
Change: 125487857
-
由 Xiaoqiang Zheng 提交于
improves by 56%: from 148 to 231 in the forward-backward pass. Support both the fastest algorithm, and fall back to the fastest algorithm without using any scratch memory, if the first algorithm fails scratch memory allocation. Soumith's conv-benchmarks stay the same before and after this change. But now it can run with bigger batch size. Change: 125484122
-
由 Dan Mané 提交于
Change: 125476357
-
由 Dan Mané 提交于
Also, if the TAG is not found, log an info rather than warning. Change: 125476271
-
由 Vijay Vasudevan 提交于
Change: 125475913
-
由 A. Unique TensorFlower 提交于
Change: 125472018
-
由 Vijay Vasudevan 提交于
Also update RELEASE.md to forward port from 0.9 branch, and add this new 'breaking change to the API.' Change: 125465285
-
由 A. Unique TensorFlower 提交于
Change: 125463524
-
由 A. Unique TensorFlower 提交于
Change: 125460867
-
由 Illia Polosukhin 提交于
Change: 125458571
-