aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
* Write common android build configs to .bazelrcGravatar Michael Case2017-10-24
| | | | | | | Add --config=android_arm and --config=android_arm64 Bazel configs to add options needed for Android Bazel builds. PiperOrigin-RevId: 173340664
* Make Iterators saveable.Gravatar Saurabh Saxena2017-10-24
| | | | | | Add tf.contrib.data.make_saveable_from_iterator(iterator) that builds a SaveableObject for an iterator so it can be saved/restored using tf.Saver. PiperOrigin-RevId: 173340191
* Add a way to run ops using a step function to MonitoredSession.Gravatar Igor Saprykin2017-10-24
| | | | | | With this method users have access to a raw Session while getting the benefit of recoverable behavior of MonitoredSession. PiperOrigin-RevId: 173334319
* Disambiguate links to "@{$estimators}".Gravatar Mark Daoust2017-10-24
| | | | PiperOrigin-RevId: 173333650
* Disables storing variables in the default variable store for eager.Gravatar Alexandre Passos2017-10-24
| | | | | | | Also disables all functional layers until a non-default store is implemented. PiperOrigin-RevId: 173333446
* Fix FusedConv2DBiasActivationOp for OIHW filter format.Gravatar A. Unique TensorFlower2017-10-24
| | | | | | The 'filter' variable wasn't initialized for OIHW filter format. PiperOrigin-RevId: 173327533
* Exposes precision_at_top_k under tf.metrics.Gravatar A. Unique TensorFlower2017-10-24
| | | | PiperOrigin-RevId: 173324359
* bool() for eager variablesGravatar Alexandre Passos2017-10-24
| | | | PiperOrigin-RevId: 173324238
* Adding the batch norm adjustment to contrib/layers.Gravatar A. Unique TensorFlower2017-10-24
| | | | PiperOrigin-RevId: 173324074
* Evaluator changes:Gravatar A. Unique TensorFlower2017-10-24
| | | | | | | | | | | * Get graph execution working, by adding init_variables() and supporting graph mode in evaluate_on_dataset(). * Use track_*() instead of add_*() to match Network. * Fill in several doc strings. * Detect metric added to two different Evaluators. * Return prefix along with metrics from metrics property. PiperOrigin-RevId: 173320585
* Removes unnecessary cast and warning in auc calculation.Gravatar A. Unique TensorFlower2017-10-24
| | | | PiperOrigin-RevId: 173320547
* Add transitional and temporary include file to mergeGravatar A. Unique TensorFlower2017-10-24
| | | | | | | grpc and grpc::internal namespaces while versions of gRPC are in transition PiperOrigin-RevId: 173317900
* Add a MultiHandler that can conditionally apply handling logic based on ↵Gravatar A. Unique TensorFlower2017-10-24
| | | | | | presence of input Tensors. PiperOrigin-RevId: 173314020
* Exception instead of crashing on resource.numpy()Gravatar Alexandre Passos2017-10-24
| | | | PiperOrigin-RevId: 173313459
* Nice error messages when using queues / batching in eager mode.Gravatar Alexandre Passos2017-10-24
| | | | PiperOrigin-RevId: 173312134
* Allow lists to be passed to tf.group().Gravatar A. Unique TensorFlower2017-10-24
| | | | PiperOrigin-RevId: 173308794
* Refactor K-FAC FisherEstimatorGravatar A. Unique TensorFlower2017-10-24
| | | | PiperOrigin-RevId: 173307212
* Eager reuse story is False instead of AUTO_REUSE.Gravatar Alexandre Passos2017-10-24
| | | | | | | | | | We want variables with eager execution to have object semantics instead of name semantics and this is a small step in that direction. This means that the functional style layer invocations (tf.layers.dense() etc.) will NOT work when eager execution is enabled. Instead, use of the object-oriented layers is advised. PiperOrigin-RevId: 173306447
* Fix NCCL rewrite bug when rerunning sessions (assigned device id is not stable).Gravatar A. Unique TensorFlower2017-10-24
| | | | | | | Fix collocate_gradients for initial losses. Remove NcclBroadcast gradient test for now. The generated AddN to accumulate the broadcast outputs before passing it to the gradient function is CPU only and cannot be collocated with NcclBroadcast on the GPU. PiperOrigin-RevId: 173306409
* replace min for std:min to avoid issues with clang compilationGravatar A. Unique TensorFlower2017-10-24
| | | | | | | e.g.: http://ci.tensorflow.org/job/nightly-matrix-linux-gpu-clang/159/console PiperOrigin-RevId: 173305545
* Better support for Metrics in graph mode:Gravatar A. Unique TensorFlower2017-10-24
| | | | | | | | * Avoid situation where variables are created in the wrong graph. * Add an init_variables() method that returns an op that will initialize any non-initialized variables. PiperOrigin-RevId: 173302832
* Fix K-FAC's loss_functions.insert_slice_in_zeros and add testsGravatar A. Unique TensorFlower2017-10-24
| | | | PiperOrigin-RevId: 173299853
* Roll forward CL 171084886Gravatar Sanjoy Das2017-10-24
| | | | | | | | | | | | | | | 171084886 had to be rolled back twice due to various open source build issues. I'm trying again, now that I think I've addressed all the pertinent issues. Original CL description: Don't use dlsym to resolve symbols in the CPU JIT Instead of resolving symbols via dlsym when JITting for the CPU backend, use a registry based mechanism. This lets us kill off the --export_dynamic hack that we used to need for CustomCall on the CPU backend. PiperOrigin-RevId: 173277862
* * Add GPU implementation of LogDeterminant op.Gravatar A. Unique TensorFlower2017-10-24
| | | | | | | | * Switch GPU implementation of Determinant to use the more numerically stable kernel as well. * Change behavior for Determinant on matrices with (numerically) infinite determinants to match the behavior of numpy.linalg.det: Return inf for matrix with infinite determinant. * Misc. cleanup in code working around missing support for complex in the NVCC compiler. PiperOrigin-RevId: 173277377
* Use tf.where instead of multiplies when masking probabilities in the ↵Gravatar A. Unique TensorFlower2017-10-24
| | | | | | BeamSearchDecoder. PiperOrigin-RevId: 173273139
* Added a dataset page to the api guideGravatar Katherine Wu2017-10-24
| | | | PiperOrigin-RevId: 173272637
* Provide better debug information on true_classes assertionGravatar A. Unique TensorFlower2017-10-24
| | | | PiperOrigin-RevId: 173266690
* [XLA] Elide whitespace in symbols in BatchNormRewriter.Gravatar Chris Leary2017-10-24
| | | | PiperOrigin-RevId: 173263867
* Creating a fix for a threading issue with ffmpeg_lib in third_party.Gravatar A. Unique TensorFlower2017-10-24
| | | | | | | | | | | | Bug is at: #5804 Fix is to add a unique identifier to each temp file name. The id is unique to the process. Multiple processes could still have a conflict, though even there the odds do go down somewhat with this fix. PiperOrigin-RevId: 173261202
* Better error message if you pass a list to tf.group().Gravatar A. Unique TensorFlower2017-10-24
| | | | PiperOrigin-RevId: 173260210
* Don't prematurely return streamsGravatar Sanjoy Das2017-10-23
| | | | PiperOrigin-RevId: 173214110
* Handle non-Layer callables and "training" arguments in Sequential.Gravatar A. Unique TensorFlower2017-10-23
| | | | | | | Create "add()" method (analogous to Keras') now that add/track_layer() is inappropriate to use. PiperOrigin-RevId: 173213945
* Internal change.Gravatar Anna R2017-10-23
| | | | PiperOrigin-RevId: 173213868
* nsync update: portability fixes for MacOS, s390x.Gravatar A. Unique TensorFlower2017-10-23
| | | | PiperOrigin-RevId: 173208878
* Updating gpu toolchainGravatar A. Unique TensorFlower2017-10-23
| | | | PiperOrigin-RevId: 173207602
* Revert to 64-bit indexing in extract_image_patches_op.h if the input/output ↵Gravatar A. Unique TensorFlower2017-10-23
| | | | | | tensor have more than 2^32 elements. PiperOrigin-RevId: 173203430
* EagerTensor.__copy__ and __deepcopy__Gravatar Alexandre Passos2017-10-23
| | | | PiperOrigin-RevId: 173194115
* Make BasicLSTMCell build its variables like a proper tf.layers.Layer.Gravatar Eugene Brevdo2017-10-23
| | | | PiperOrigin-RevId: 173193748
* Adds the ability to adjust the normalized value in batchnorm. By using ↵Gravatar A. Unique TensorFlower2017-10-23
| | | | | | random adjustments, it is possible to improve the model generalization. PiperOrigin-RevId: 173192179
* tf.confusion_matrix works with eager execution enabled.Gravatar Alexandre Passos2017-10-23
| | | | PiperOrigin-RevId: 173188867
* Fix error in topk heap launch code. It assumed sizeof(struct) == sizeof(structGravatar A. Unique TensorFlower2017-10-23
| | | | | | components). PiperOrigin-RevId: 173188044
* tf.constant takes numpy dtypes in eager mode as wellGravatar Alexandre Passos2017-10-23
| | | | PiperOrigin-RevId: 173184568
* metrics.Mean writes a summary.Gravatar Alexandre Passos2017-10-23
| | | | PiperOrigin-RevId: 173178780
* Automated g4 rollback of changelist 172946149Gravatar Sanjoy Das2017-10-23
| | | | PiperOrigin-RevId: 173176850
* [XLA] Adds a C64 type to XLA, with actual compilation support coming soon.Gravatar A. Unique TensorFlower2017-10-23
| | | | PiperOrigin-RevId: 173172916
* Infer `tf.contrib.distributions.RelaxedOneHotCategorical` `dtype` from ↵Gravatar A. Unique TensorFlower2017-10-23
| | | | | | arguments. PiperOrigin-RevId: 173172808
* Make Optimizer.minimize work when eager execution is enabled.Gravatar A. Unique TensorFlower2017-10-23
| | | | PiperOrigin-RevId: 173172604
* Java: Update release to 1.4.0-rc1Gravatar Asim Shankar2017-10-23
| | | | PiperOrigin-RevId: 173172018
* Remove name_scope from convolutional calls.Gravatar A. Unique TensorFlower2017-10-23
| | | | PiperOrigin-RevId: 173171871
* Makes gradients_function exception-safe.Gravatar Alexandre Passos2017-10-23
| | | | PiperOrigin-RevId: 173170394
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-04 06:03:20 +0000