- 08 4月, 2016 23 次提交
-
-
由 Manjunath Kudlur 提交于
by all Envs. Change: 119342847
-
由 Manjunath Kudlur 提交于
Change: 119338733
-
由 A. Unique TensorFlower 提交于
Change: 119336366
-
由 Derek Murray 提交于
This is a breaking change! The following classes have been renamed: tf.GrpcServer -> tf.train.Server tf.ClusterSpec -> tf.train.ClusterSpec tf.ServerDef -> tf.train.ServerDef tf.JobDef -> tf.train.JobDef tf.ClusterDef -> tf.train.ClusterDef The constructor for tf.train.Server is more permissive and now accepts tf.train.ClusterSpec, tf.train.ClusterDef, and dictionary inputs for specifying the cluster part of the server. For consistency, the server library moves from python/client to python/training. Change: 119335624
-
由 A. Unique TensorFlower 提交于
Change: 119325959
-
由 A. Unique TensorFlower 提交于
Change: 119324250
-
由 A. Unique TensorFlower 提交于
Add AUC Op based on internal histograms: auc_using_histogram(). * Works fine with large class imbalance (no special handling when entire batch is one class) * Runs in O(batch_size + histogram_width) time, so works for huge batches Change: 119324088
-
由 A. Unique TensorFlower 提交于
Change: 119323528
-
由 A. Unique TensorFlower 提交于
Change: 119322379
-
由 A. Unique TensorFlower 提交于
Change: 119322171
-
由 Sherry Moore 提交于
Change: 119321281
-
由 A. Unique TensorFlower 提交于
Change: 119320795
-
由 Kiril Gorovoy 提交于
Fix genrules that didn't work when TensorFlow was imported as a submodule and compiled with --config=cuda. Change: 119318629
-
由 Sherry Moore 提交于
Change: 119313512
-
由 A. Unique TensorFlower 提交于
Change: 119313360
-
由 Josh Levenberg 提交于
Change: 119312877
-
由 A. Unique TensorFlower 提交于
This is useful for multi-task setups where a network is trained on one task and then a subset of the layers are reused for a different (but related) task. Change: 119310581
-
由 A. Unique TensorFlower 提交于
Change: 119302560
-
由 A. Unique TensorFlower 提交于
If TF_CUDA_COMPUTE_CAPABILITIES version is not set and deviceQuery is available, use the version extracted from deviceQuery output. Change: 119292986
-
由 Yuan Yu 提交于
Change: 119288985
-
由 A. Unique TensorFlower 提交于
Change: 119284140
-
由 Benoit Steiner 提交于
and polygamma functions, as well as improved support for float16. Change: 119279101
-
由 Dan Smilkov 提交于
Change: 119276088
-
- 07 4月, 2016 15 次提交
-
-
由 A. Unique TensorFlower 提交于
Change: 119270996
-
由 Sherry Moore 提交于
Change: 119269792
-
由 Manjunath Kudlur 提交于
- Env dispatches to a FileSystem interface - FileSystemFactory is used to look up the correct FileSystem implementation based on the prefix of the filename - Provide a registration mechanism to register different factories Change: 119268846
-
由 A. Unique TensorFlower 提交于
Change: 119237059
-
由 Manjunath Kudlur 提交于
- Env dispatches to a FileSystem interface - FileSystemFactory is used to look up the correct FileSystem implementation based on the prefix of the filename - Provide a registration mechanism to register different factories Change: 119236345
-
由 A. Unique TensorFlower 提交于
Change: 119234421
-
由 Manjunath Kudlur 提交于
- Env dispatches to a FileSystem interface - FileSystemFactory is used to look up the correct FileSystem implementation based on the prefix of the filename - Provide a registration mechanism to register different factories Change: 119233977
-
由 A. Unique TensorFlower 提交于
Change: 119221959
-
由 A. Unique TensorFlower 提交于
Change: 119213211
-
由 A. Unique TensorFlower 提交于
Change: 119212533
-
由 Sherry Moore 提交于
Change: 119196706
-
由 A. Unique TensorFlower 提交于
Change: 119194215
-
由 A. Unique TensorFlower 提交于
when I converted ops and added unit tests.) Change: 119189211
-
由 A. Unique TensorFlower 提交于
Change: 119171821
-
由 A. Unique TensorFlower 提交于
numbers.h. This removes the duplicate code, but has a small behavior change - safe_strto32 strips leading and trailing whitespace, while NumericParse does not. But this seems okay. Also changed the flag parsing command_line_flags.cc to template on int32 instead of int. Change: 119170865
-
- 06 4月, 2016 2 次提交
-
-
由 A. Unique TensorFlower 提交于
Change: 119119175
-
由 A. Unique TensorFlower 提交于
Change: 119104425
-