aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
...
* Add debug output to CHECK for compatible shapes of multi-output fusions.Gravatar A. Unique TensorFlower2018-06-04
| | | | PiperOrigin-RevId: 199091580
* Go: Update generated wrapper functions for TensorFlow ops.Gravatar A. Unique TensorFlower2018-06-03
| | | | PiperOrigin-RevId: 199072157
* Update ops-related pbtxt files.Gravatar A. Unique TensorFlower2018-06-03
| | | | PiperOrigin-RevId: 199071075
* [tf.data] Input pipeline rewrites prototype.Gravatar Jiri Simsa2018-06-03
| | | | | | | | This CL: - adds `tf.contrib.data.optimize()` transformation that can be used to trigger rewrite-based optimization for the input pipeline. - adds `tf.data.Dataset._as_serialized_graph()` method that returns the serialized graph representation of the dataset PiperOrigin-RevId: 199068055
* New NN API interface that uses the TensorFlow Lite delegate API.Gravatar Andrew Selle2018-06-03
| | | | | | | | | - Make nn_api a delegate in its own directory. - Use the delegate API to rewrite the graph. - Use only on static APIs right now. - This is initial preview of the delegate that only supports add and conv. PiperOrigin-RevId: 199055747
* Don't cluster Identity nodes that forward tensor refsGravatar Sanjoy Das2018-06-02
| | | | | | | XLA cannot implement the forward-tensor-ref semantic -- there is no guaranteed aliasing between the input and output of the XLA cluster. PiperOrigin-RevId: 199005227
* Adding support for the int() and float() built-ins.Gravatar A. Unique TensorFlower2018-06-02
| | | | PiperOrigin-RevId: 199001807
* Replace boilerplate code with function template.Gravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198963930
* Quantize weights transformation for toco.Gravatar Suharsh Sivakumar2018-06-01
| | | | | | Finds float weight tensors, quantizes them to 8 bits, and adds Dequantize operations after them. PiperOrigin-RevId: 198955123
* [XLA] Add comments for the Reduce->Reshape simplifier pass.Gravatar Kay Zhu2018-06-01
| | | | | | Also forcing reduction order for init to be on lhs for ReduceWindow->Map pass. PiperOrigin-RevId: 198953817
* Extract FoldMultiplyIntoConv optimization stage.Gravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198953044
* Remove use of absl::make_uniqueGravatar Brennan Saeta2018-06-01
| | | | | | absl is not yet ready for use by open source TensorFlow. :-( PiperOrigin-RevId: 198952953
* Add an explanatory comment.Gravatar Sanjoy Das2018-06-01
| | | | PiperOrigin-RevId: 198949796
* Internal change.Gravatar Anna R2018-06-01
| | | | PiperOrigin-RevId: 198948296
* Change the Eigen reduction code to use a tree to improve numerical stability.Gravatar A. Unique TensorFlower2018-06-01
| | | | | | | | | | | | | | This changes the InnerMostDimReducer to use a summation tree, which is more numerically stable than the previous approach of sequential addition into an accumulator. This solves the issue for reduction over all or a trailing subset of dimensions. This change does not improve the numerical accuracy for MeanReducer, which maintains state. Benchmarks show a 40% (AVX) to 50% (SSE) slowdown for small row reductions (sum, float). column- and full reductions are unchanged. Cleaned up TensorFunctors.h a bit by moving the traits to reducer_traits and updating the code that uses the reducers accordingly. Introduced a new trait "IsExactlyAssociative" and new template specializations of InnerMostDimReducer to ensure that we only invoke the new and slightly more expensive codepath when it is needed, i.e. for sum reduction of non-integer types. PiperOrigin-RevId: 198946075
* Allow user to opt out of saving metagraph for TPU with ↵Gravatar A. Unique TensorFlower2018-06-01
| | | | | | TPUEstimator.export_output(). PiperOrigin-RevId: 198944144
* Support 5-inputs LSTM kernel in TFLite (float only).Gravatar Yu-Cheng Ling2018-06-01
| | | | PiperOrigin-RevId: 198943559
* Go: Update generated wrapper functions for TensorFlow ops.Gravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198942995
* Update ops-related pbtxt files.Gravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198941362
* Adds a batch-op implemented using TF functions.Gravatar Vinu Rajashekhar2018-06-01
| | | | | | | | o This has a couple of important advantages over the current implementation: 1. The existing batch-op waits for the batch to be created and then forwards the tensors to the rest of the graph, which causes a lot of batches to be created, because there is no way for the op to know if the other batches are being queued up. A mitigation, which we have seen working in practice, is to actually wait for the graph to finish processing the batch. So there is a sort of flow-control happening, and meanwhile the batches get coalesced, which improves latency and throughput as well. Using functions makes this kind of approach easier. 2. The existing op passes empty tensors around the graph to make the TF executor happy, which has sometimes worked not well with some Ops (like Reshape). Using functions means that we don't need to rely on this mechanism as well. PiperOrigin-RevId: 198937594
* Swift for TensorFlow lives in GitHub, for now.Gravatar Billy Lamberta2018-06-01
| | | | | | | Update ecosystem page and dropdown menu. Remove community/swift page and add redirect. PiperOrigin-RevId: 198936463
* [XLA] Move xla/tools/parser/* into xla/service.Gravatar Justin Lebar2018-06-01
| | | | | | | | Now that we're using the parser inside of xla/service, it's awkward for it to live inside of xla/tools, because everything else in there is a standalone tool. We've already had one person be confused by this. PiperOrigin-RevId: 198935921
* DepthwiseConv OptimizationsGravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198935499
* [XLA] Add an unoptimized HLO output flag to ExecutableBuildOptions and to ↵Gravatar Roy Frostig2018-06-01
| | | | | | the XLA local Python client. PiperOrigin-RevId: 198930874
* Optimized implementation of dilated convolution. Added a DilatedIm2Col() ↵Gravatar A. Unique TensorFlower2018-06-01
| | | | | | function to leverage GEMM optimizations. PiperOrigin-RevId: 198924313
* [xla] expose a ConvGeneralDilated op in the local Python clientGravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198922037
* [TF:XLA] Bump open source llvm revision to r333732Gravatar Sanjoy Das2018-06-01
| | | | PiperOrigin-RevId: 198921960
* In the Swift API, deprecate `a.dot(b)` and `?` to `matmul(a, b)` to ↵Gravatar Richard Wei2018-06-01
| | | | | | | | | | accurately reflect the operator?s mathematical properties and make it familiar to TensorFlow users. Currently the deprecation is a warning - when we update tensorflow/swift-models, I'll start another CL to remove it completely. Previously `dot` was chosen over `matmul` because of naming convention concerns (acronyms aren?t common in Swift) and that we wanted to make it short (so full names like `a.matrixMultiplied(by: b)` isn?t acceptable). Beyond these concerns, `matmul` is really a word of art and thus should be preferred. The ? operator often denotes outer product and Kronecker product. So it's removed, too. PiperOrigin-RevId: 198920621
* Add wrapper header file for SerialDeviceBatchSchedulerGravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198919964
* TFLite: adding tile and expand_dims ops.Gravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198913026
* [TF:XLA] Refactor implementation of TruncatedNormal to avoid redundant ↵Gravatar Peter Hawkins2018-06-01
| | | | | | computations. Add an additional test. PiperOrigin-RevId: 198908904
* [TF:XLA] Bump open source llvm revision to r333578Gravatar Sanjoy Das2018-06-01
| | | | PiperOrigin-RevId: 198906281
* [TF2XLA] Decompose resize bilinear with large filters to work on dimensionsGravatar Blake Hechtman2018-06-01
| | | | | | indpendently. PiperOrigin-RevId: 198902279
* Printing bools in graphviz.Gravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198897530
* Amend cluster resolver error to suggest oauth2client as a possible issue.Gravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198894470
* Updates Interpreter to be initialized with a MappedByteBuffer for backward ↵Gravatar A. Unique TensorFlower2018-06-01
| | | | | | compatibility. PiperOrigin-RevId: 198893078
* Fix bug in eager documentation.Gravatar Akshay Agrawal2018-06-01
| | | | | | | When implementing a custom layer, it's necessary to call the Layer constructor from the custom layer's constructor. PiperOrigin-RevId: 198892503
* Add a dependency optimization that eliminates multiple cross-device control ↵Gravatar Peter Hawkins2018-06-01
| | | | | | edges to a single node from the same source device. Instead, build an intermediate NoOp node on the source device and use a single cross-device control edge. PiperOrigin-RevId: 198891614
* Fix ProfileSummarizer build, use properly qualified string references.Gravatar Shashi Shekhar2018-06-01
| | | | PiperOrigin-RevId: 198887868
* Fixed a bug introduced by cl/197941474.Gravatar Benoit Steiner2018-06-01
| | | | PiperOrigin-RevId: 198886485
* remove typoGravatar Gunhan Gulsoy2018-06-01
| | | | PiperOrigin-RevId: 198880096
* Automated g4 rollback of changelist 198815200Gravatar Rachel Lim2018-06-01
| | | | PiperOrigin-RevId: 198878259
* Automated g4 rollback of changelist 198810875Gravatar Peter Hawkins2018-06-01
| | | | PiperOrigin-RevId: 198876135
* implement a generic reduce method so that later we can easily implement ↵Gravatar A. Unique TensorFlower2018-06-01
| | | | | | reduce_{sum,prod,etc} PiperOrigin-RevId: 198874465
* Resubmitting CL 196349902: Adding cuDNN header dependency to targets that ↵Gravatar A. Unique TensorFlower2018-06-01
| | | | | | include the cuDNN header file. PiperOrigin-RevId: 198869605
* Disable test on windows until we figure out what's wrong.Gravatar Dan Moldovan2018-06-01
| | | | PiperOrigin-RevId: 198863091
* Mark tensorflow/python/kernel_tests/linalg:linear_operator_identity_test as ↵Gravatar Peter Hawkins2018-06-01
| | | | | | optonly due to flakiness. PiperOrigin-RevId: 198862313
* Support bfloat16 in LiteralBase::SliceGravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198859282
* Fix a bug for unspecified dtype of acc_shape that can cause type mismatch.Gravatar A. Unique TensorFlower2018-06-01
| | | | PiperOrigin-RevId: 198850955
* Use ConstantDataArray to lower arrays of constants.Gravatar A. Unique TensorFlower2018-06-01
| | | | | | For large constants, creating an llvm::Constant for each element can get prohibitively large compile times. PiperOrigin-RevId: 198843141
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-06 15:41:26 +0000