aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
* variables_to_restore: Differentiate python variables by string name rather ↵Gravatar Suharsh Sivakumar2017-11-07
| | | | | | | | | | than object. variables_to_restore ensured that duplicate variables weren't added to the return map by comparing python variable object. Normally there is only one Variable object for each underlying variable, so this wasn't a problem. But when one initializes a graph by importing a GraphDef, duplicate python Variable objects are created for each occurrence of a variable in a collection (say, global variables and moving average variables). This change fixes variables_to_restore to work with an imported graph def by not comparing Variable objects. PiperOrigin-RevId: 174861804
* Added CPU matrix exponential op to TensorFlow.Gravatar A. Unique TensorFlower2017-11-07
| | | | | | Uses Eigen's unsupported implementation. PiperOrigin-RevId: 174858966
* Use nesting to reduce the number of modules listed in the API TOC.Gravatar Mark Daoust2017-11-07
| | | | PiperOrigin-RevId: 174846842
* Previously if ComputeConstant seen a parameter it failed to proceed.Gravatar A. Unique TensorFlower2017-11-07
| | | | | | | | | | | After this change we can specify a list of parameters to it and if we specify enough then it will do the computation. The primary goal of this change is to make the HloEvaluator usable with ComputationBuilder from tests through ComputeConstant in cases where the input is a parameter (fed by a literal). PiperOrigin-RevId: 174845108
* Add a unit test for gradient computation with layout optimizer.Gravatar Yao Zhang2017-11-06
| | | | PiperOrigin-RevId: 174814136
* Refactor function Optimize.Gravatar Yao Zhang2017-11-06
| | | | PiperOrigin-RevId: 174813300
* [TF:XLA] Add two disabled tests with while ops that permute tuple elements.Gravatar A. Unique TensorFlower2017-11-06
| | | | | | | | | | These tests permute the tuple elements of a 3-tuple in each iteration in the following cyclic manner (132), i.e. a shift to the left. The first test just return the result tuple, the second returns the sum of all tuple elements (which is expected to be constant 6, no matter which permutation) Both tests are disabled for now because they fail on all back-ends. PiperOrigin-RevId: 174806092
* Fix for LLVM API changes for fast math (https://reviews.llvm.org/rL317488).Gravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174799735
* Add InceptionResNetV2 to tf.keras and update applications module to match ↵Gravatar Francois Chollet2017-11-06
| | | | | | Keras 2.0.9. PiperOrigin-RevId: 174796893
* Automated g4 rollback of changelist 174735029Gravatar Guangda Lai2017-11-06
| | | | PiperOrigin-RevId: 174796480
* Skip non-existent fetch nodesGravatar Benoit Steiner2017-11-06
| | | | PiperOrigin-RevId: 174795864
* Automated g4 rollback of changelist 174523638Gravatar Justin Lebar2017-11-06
| | | | PiperOrigin-RevId: 174788331
* Update ops-related pbtxt files.Gravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174787397
* Remove binary dependency from optimize_for_inference_libGravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174787363
* [TF TensorArray] allow reading from an unwritten index if fully defined ↵Gravatar Eugene Brevdo2017-11-06
| | | | | | | | | element_shape is given. This allows one to write to only some indices of a TensorArray before calling stack. Elements that were not written to are treated as all zero tensors. PiperOrigin-RevId: 174783569
* [BufferAssignment] Sort allocation's "Assigned" objects before converting to ↵Gravatar Yunxing Dai2017-11-06
| | | | | | | | | a proto. This makes the buffer assignment's proto dump deterministic. RELNOTES: BufferAssignment's protocol buffer dump is now deterministic. PiperOrigin-RevId: 174783549
* Go: Update generated wrapper functions for TensorFlow ops.Gravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174781987
* Allow for an old_row_vocab_size, in case a subset of the old_row_vocab_file ↵Gravatar A. Unique TensorFlower2017-11-06
| | | | | | was used during the checkpoint creation (as is allowed in FeatureColumn._VocabularyListCategoricalColumn). PiperOrigin-RevId: 174781749
* Fix DepthToSpaceGrad and SpaceToDepthGrad on data_format NCHW.Gravatar Jingyue Wu2017-11-06
| | | | | | This fixes #14243. PiperOrigin-RevId: 174772870
* Add `parallel_interleave` to tf/contrib/data/__init__.py so that it is ↵Gravatar Frank Chen2017-11-06
| | | | | | directly addressable from tf.contrib.data. PiperOrigin-RevId: 174771870
* Refactor eager benchmarks to subclass Benchmark.Gravatar Akshay Agrawal2017-11-06
| | | | PiperOrigin-RevId: 174770787
* Supports multi-dimensional logits and labels in multi class head.Gravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174770444
* Fix race bug in AdaptiveSharedBatchScheduler.Gravatar A. Unique TensorFlower2017-11-06
| | | | | | In ASBSQueue::Schedule, when a new batch is created, it was added to the scheduler outside of the queue's lock. This was done to prevent any unforeseen interactions between the queue lock and scheduler lock. However, this wasn't being done in a thread safe way. PiperOrigin-RevId: 174769383
* Fix device querying in Keras backend.Gravatar Francois Chollet2017-11-06
| | | | PiperOrigin-RevId: 174769308
* More thoroughly disable the should_use_result decorator when executing eagerly.Gravatar Allen Lavoie2017-11-06
| | | | | | | | It was creating reference cycles. Adds a test that TensorArrays create no reference cycles in eager mode. PiperOrigin-RevId: 174768765
* Remove use of xrange which is not python3 compatible.Gravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174768741
* Force CUDA runtime initialization only when device count is larger than 0.Gravatar Yangzihao Wang2017-11-06
| | | | PiperOrigin-RevId: 174767565
* Create a routine that can collapse a subgraph into a fused opGravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174765540
* BUGFIX: special_math.ndtri didn't work with dynamic shapes. This was due ↵Gravatar Ian Langmore2017-11-06
| | | | | | to use of constant_op.constant(..., shape=p.shape), where sometimes p was a Tensor of unknown shape. PiperOrigin-RevId: 174764744
* Fold Transpose into Matmul and SparseMatmul.Gravatar A. Unique TensorFlower2017-11-06
| | | | | | Fold ConjugateTranspose in BatchMatmul. PiperOrigin-RevId: 174750173
* Replace the implementation of tf.flags with absl.flags.Gravatar Yilei Yang2017-11-06
| | | | | | | | | | | | Previous tf.flags implementation is based on argparse. It contains -h/--help flags, which displays all flags. absl.app's --help flag only displays flags defined in the main module. There is a --helpfull flag that displays all flags. So added --helpshort --helpfull flags. app.run now raises SystemError on unknown flags (fixes #11195). Accessing flags before flags are parsed will now raise an UnparsedFlagAccessError, instead of causing implicit flag parsing previously. PiperOrigin-RevId: 174747028
* Fix invalid flush_secs argument.Gravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174745329
* Refactor helper functions a bit for virtual gpu changes later.Gravatar Guangda Lai2017-11-06
| | | | PiperOrigin-RevId: 174735029
* [tf.data] Remove unused members from IteratorContext.Gravatar Derek Murray2017-11-06
| | | | PiperOrigin-RevId: 174734277
* Improved encoding on shapes in grappler.Gravatar Benoit Steiner2017-11-06
| | | | PiperOrigin-RevId: 174733491
* Clean up some reference cycles in eager mode.Gravatar Allen Lavoie2017-11-06
| | | | | | | ResourceVariables enter graph mode to get a handle. We should probably revisit that, but in the meantime we can break the resulting reference cycles. PiperOrigin-RevId: 174732964
* [tpu:profiler] Support the Input Pipeline Analyzer tool in TPU profiler (WIP)Gravatar A. Unique TensorFlower2017-11-06
| | | | | | | o. move input pipeline analyzer related proto for grpc between red and green VMs o. rename perftools.gputools.profiler.collector::TfStatsHelperResult to tensorflow::tpu::TfOpStats. PiperOrigin-RevId: 174730411
* Memory leak detected when running a heap checker in our tests.Gravatar Alexandre Passos2017-11-06
| | | | PiperOrigin-RevId: 174726228
* Basic plumbing for calling C API from import_graph_def()Gravatar Skye Wanderman-Milne2017-11-06
| | | | PiperOrigin-RevId: 174724070
* Remove misleading comment from Eigen build file.Gravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174719222
* Add input format agnostic way to parse HLOs.Gravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174719153
* Fix generate_examples buildGravatar Andrew Selle2017-11-06
| | | | | | | | | | | | | | - Add -march=native to host_copts and host_cxxopts in configure.py - Make string.h for abstracting string differences at core interpreter level - Use tensorflow special arg parse instead of flags - Switch to using tool instead of data for dependency - Fix python3 compatibility + Use six.StringIO instead of StringIO.StringIO + Use print_function + Properly set binary flags on TempFile's used in toco_convert - Misc other path fixes PiperOrigin-RevId: 174717673
* Improve error message when a function is already defined with the same name ↵Gravatar RJ Ryan2017-11-06
| | | | | | and different hash string. PiperOrigin-RevId: 174715563
* SerializeIterator: do not unref the resource until we're finished using it.Gravatar Derek Murray2017-11-06
| | | | | | This change avoids a potential use-after-free error if the resource is concurrently serialized and destroyed (e.g. by a DestroyResourceOp or Session::Reset()). PiperOrigin-RevId: 174713115
* Enables wrapping input pipeline into tf.while_loop for all users.Gravatar Jianwei Xie2017-11-06
| | | | PiperOrigin-RevId: 174708213
* Replicate `Estimator.model_fn` across available GPUs.Gravatar Igor Saprykin2017-11-06
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | def replicate_model_fn(model_fn, optimizer_fn, devices=None): """Replicate `Estimator.model_fn` over GPUs. ... I tested that it seems to give the right result on cnn_mnist.py on 1 CPU, 1 real GPU, 4 allow_soft_placement=True GPUs. Some measurements on CNN MNIST across steps 19300-20000: 1) no replicate_model_fn call: global_step/sec: 156.254 global_step/sec: 155.074 global_step/sec: 155.74 global_step/sec: 153.636 global_step/sec: 157.218 global_step/sec: 159.644 2) replicate across one hardware GPU: global_step/sec: 158.171 global_step/sec: 165.618 global_step/sec: 162.773 global_step/sec: 159.204 global_step/sec: 162.289 global_step/sec: 167.173 3) replicate across 4 software GPUs on one hardware GPU (soft placement): global_step/sec: 75.47 global_step/sec: 76.16 global_step/sec: 75.18 Loss numbers didn't change across the three configurations. PiperOrigin-RevId: 174704385
* Add profiling support to Service::ExecuteParallel.Gravatar A. Unique TensorFlower2017-11-06
| | | | PiperOrigin-RevId: 174682772
* Use latest nsync: allows running bazel after having downloaded for "make" buildGravatar A. Unique TensorFlower2017-11-05
| | | | | | | | | The downloads directory for the make build is within the source tree seen by bazel, which means that BUILD files (by whatever name) without those downloaded trees must all be valid in their new location, or not recognized by bazel as being BUILD files. The new version of nsync handles that, and this change pulls in that new version. PiperOrigin-RevId: 174652898
* [TF2XLA] Don't change output port for control dependency in CopySubgraph.Gravatar Jacques Pienaar2017-11-05
| | | | | | | If the output is being squashed then we want control output 0, except where the input is a control dependency. PiperOrigin-RevId: 174633829
* Add tf.contrib.framework.sort, wrapping tf.nn.top_k (#288).Gravatar Dan Ringwalt2017-11-04
| | | | | | Comparable to np.sort, but their "kind" parameter is not implemented (only one sort algorithm) and "order" is not applicable (tensors do not have fields). PiperOrigin-RevId: 174588000
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-05 05:00:22 +0000