aboutsummaryrefslogtreecommitdiffhomepage
Commit message (Collapse)AuthorAge
...
| | | | * BREAKING_CHANGE: Split out event_ndims=0 bijectors from Affine and ↵Gravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | CholeskyOuterProduct. - Deprecate event_ndims argument - Create a Square bijector for the scalar case of CholeskyOuterProduct (which now only operates on matrices). - Create a AffineScalar bijector for the scalar case of Affine (which now only operates on vectors) PiperOrigin-RevId: 188801116
| | | | * Switch BuildGraphOptions to wrap CallableOptions.Gravatar Derek Murray2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This change harmonizes the graph construction codepaths for DirectSession and MasterSession, which will make it easier to add new subgraph creation features. PiperOrigin-RevId: 188799932
| | * | | fix build file via buildifierGravatar Frank Chen2018-03-12
| | | | |
| | | | * Don't use shapes unless it's safe to do so.Gravatar Benoit Steiner2018-03-12
| | | | | | | | | | | | | | | | | | | | PiperOrigin-RevId: 188796626
| | | | * [tf.data] Enable Dataset.make_one_shot_iterator() and Dataset.__iter__() in ↵Gravatar Derek Murray2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | eager mode. This change partially replicates the code in `tf.contrib.eager.Iterator`. However, since that class depends on contrib-level functionality (viz. cross-device prefetching support), we cannot move it wholesale to core. PiperOrigin-RevId: 188790349
| | | * | Fix the script entry point for freeze_graph.Gravatar Mark Daoust2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The wrapper created by `setup.py` calls the entry point function with no arguments. `freeze_graph.main` expects the global `FLAGS` to be set, and one argument. This change adds a `run_main` function to use as the entry point, which expects no arguments and parses the flags. It also adds a `flags` argument to `main` so the flags can be passed directly without using a `global FLAGS` declaration.
* | | | | Enable CUDNN_TENSOR_OP_MATH for fp16 RNNs (#17367)Gravatar Ben Barsdell2018-03-12
| |_|/ / |/| | | | | | | | | | | | | | | | | | | | | | | - Speeds up CUDNN RNNs with fp16 input/output when possible on supported GPUs. Computations will fall back to pseudo-fp16 if tensor op math is not supported. - Enabled by default, but can be disabled by setting the environment variable TF_DISABLE_CUDNN_RNN_TENSOR_OP_MATH=1.
| | | * TFLite Delegate: Expose input / output tensor indicies in `Init`Gravatar Yu-Cheng Ling2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188784614
| | | * Fix typo in the doc: tf.multiplytiplyGravatar Max Galkin2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | https://www.tensorflow.org/api_docs/python/tf/multiply PiperOrigin-RevId: 188782466
| | | * K-FAC: FisherBlocks for tf.nn.{depthwise_conv2d, separable_conv2d, convolution}.Gravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188778072
| | | * Move "hoist common factor out of aggregation" optimizationGravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | to a separate stage. 1) Use a new naming scheme for optimized ops, share it with AddOpsRewrite 2) Make sure that tests actually test that optimized nodes exists in a graph PiperOrigin-RevId: 188772892
| | | * [TF:XLA] Bump open source llvm revision to r327201Gravatar Sanjoy Das2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188771994
| | | * [TFLite] Don't require a std::vector for Interpreter::SetTensorParameters*.Gravatar Eugene Brevdo2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188770522
| | | * Move `loss_reduction` argument from `replicate_model_fn` to `TowerOptimizer.Gravatar Igor Saprykin2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188766477
| | | * Add is_discrete, is_continuous, is_bounded methods to TensorSpecs.Gravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188766232
| | | * Fix another eager PyObject leakGravatar Allen Lavoie2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | Shockingly this one was also due to PySequence_GetItem. PiperOrigin-RevId: 188765548
| | | * Add custom_gradient function.Gravatar Joshua V. Dillon2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188765271
| | | * Extend TF Eager C API to allow asynchronous execution.Gravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188763442
| | | * [XLA] [Copy insertion] Deterministically iterate through instructions to copyGravatar Yunxing Dai2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | - Use HloInstructionMap to get deterministic iteration order. PiperOrigin-RevId: 188755375
| | | * Switch op_hint.py to use _set_attr.Gravatar Skye Wanderman-Milne2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | This is in preparation for enabling the C API. Modifying an op's NodeDef directly has no effect with the C API enabled. PiperOrigin-RevId: 188754464
| | | * Standardize "op" capitalization, see "adding_an_op".Gravatar Mark Daoust2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188753529
| | | * Improve usability of `tf.contrib.bayesflow.custom_gradient` by removing need ↵Gravatar Joshua V. Dillon2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | for `axis` arg and support taking lists. PiperOrigin-RevId: 188751894
| | | * Transposes are can be merged into reshapes when the ordering of non-oneGravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | dimensions remains unchanged. PiperOrigin-RevId: 188751074
| | | * boosted_trees: infer the output shapes of Quantiles Op from the input shapes.Gravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188750079
| | | * Avoid capturing unused variables in lambda functionsGravatar Benoit Steiner2018-03-12
| | | | | | | | | | | | | | | | PiperOrigin-RevId: 188747641
| | | * Fix race in C API.Gravatar Skye Wanderman-Milne2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | RecordMutation could race with ExtendSessionGraphHelper, which would release the graph lock and only keep the session lock when extending the session. Also makes sure thread annotations are on declarations, not definitions (otherwise they have no effect). PiperOrigin-RevId: 188747158
| | * | Fix do_check_load_py_testGravatar Frank Chen2018-03-12
| | | |
| | * | More clean-upGravatar Frank Chen2018-03-12
| | | |
| | | * Turn trivial Pack ops with a single input into ExpandDims ops to avoid ↵Gravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | copying the tensor. PiperOrigin-RevId: 188742516
| | * | Clean up BUILD fileGravatar Frank Chen2018-03-12
| | | |
| | | * Make default number of threads trigger the default behavior for both eigenGravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | and gemmlowp. In gemmlowp the default is '1', while in eigen it is 'number of processors'. PiperOrigin-RevId: 188742087
| | * | Branch 188738133Gravatar Frank Chen2018-03-12
| |/| | |/| | |
| | | * Convert Squeeze into Reshape: Support empty output shapes.Gravatar A. Unique TensorFlower2018-03-12
| | |/ | | | | | | | | | PiperOrigin-RevId: 188740288
| | * Supporting quantization of Gather ops and removal of trivial Relu1s when ↵Gravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | | | quantized. PiperOrigin-RevId: 188738133
| | * Plug a few more PyObject leaks, test for them.Gravatar Allen Lavoie2018-03-12
| | | | | | | | | | | | PiperOrigin-RevId: 188731961
| | * Don't let the grappler item builder fail if the graph contains unknown customGravatar Benoit Steiner2018-03-12
| | | | | | | | | | | | | | | | | | ops. PiperOrigin-RevId: 188730560
| | * Don't remove identity nodes if they follow a device crossing and have ↵Gravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | consumers on a device different than themselves. They may be used to cache or route data between devices in a deliberate manner. Simplify code in DependencyOptimizer a bit. PiperOrigin-RevId: 188730185
| | * ResourceScatterUpdate: Gracefully handle inconsistent indices and updates inGravatar Asim Shankar2018-03-12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | the kernel. With graph execution, consistency between the shapes of the arguments to ResourceScatterUpdate is validated by the shape inference functions at graph construction time. With eager execution, the shape inference logic isn't executed, so inconsistent arguments could be provided to the kernel, which would result in a segmentation fault prior to this change. As demonstrated by the added tests. PiperOrigin-RevId: 188729154
| | * Turn on function optimization by defaultGravatar Benoit Steiner2018-03-12
| | | | | | | | | | | | PiperOrigin-RevId: 188722505
| | * Reuse the linear index when broadcasting a contiguous range of dimensions.Gravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | | | | | | | This potentially allows us to get rid of additional mod and div operations. PiperOrigin-RevId: 188719238
| | * Lint some files.Gravatar A. Unique TensorFlower2018-03-12
| | | | | | | | | | | | PiperOrigin-RevId: 188698275
| | * this test is also timing out in cuda so disabling for nowGravatar Olivia Nordquist2018-03-12
| | | | | | | | | | | | PiperOrigin-RevId: 188685611
| | * propagate fix from tensorflow/models#3561Gravatar Mark Daoust2018-03-11
| | | | | | | | | | | | PiperOrigin-RevId: 188675327
| | * Fix assets for the TF camera example.Gravatar Shashi Shekhar2018-03-11
| | | | | | | | | | | | | | | | | | Mobile net model is downloaded from tf_http_archive("tf_mobilenet") rule and renaming the asset file in assets folder has no effect. PiperOrigin-RevId: 188672531
| | * disable flaky asan testGravatar Olivia Nordquist2018-03-11
| | | | | | | | | | | | PiperOrigin-RevId: 188670616
| | * Fix typo in description of INTERNAL error code.Gravatar Derek Murray2018-03-11
| | | | | | | | | | | | PiperOrigin-RevId: 188666142
| | * Removed duplicate statement.Gravatar A. Unique TensorFlower2018-03-11
| | | | | | | | | | | | PiperOrigin-RevId: 188663018
| | * Fixes a race condition in function instantiation.Gravatar Derek Murray2018-03-11
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Previously, if the same function was being concurrently instantiated and released: 1. Thread one could begin to instantiate the function, determine that it already existed in the runtime, then be preempted. 2. Thread two could release the handle on the function, causing it to be freed and removed from the `FunctionLibraryRuntime::items_` map. 3. Thread one could then incorrectly assume that the function still existed, and fail to find it in the `FunctionLibraryRuntime::items_` map, causing a segfault when it attempted to increment the refcount on an uninitialized object. PiperOrigin-RevId: 188661500
| | * Specify the `maximum_iterations` to tf.while_loop in tf.scan to be ↵Gravatar A. Unique TensorFlower2018-03-11
| | | | | | | | | | | | | | | | | | compatible with XLA. PiperOrigin-RevId: 188652533
| | * Selectively re-enable bfloat16 tests for the GPU backend.Gravatar Bixia Zheng2018-03-11
| | | | | | | | | | | | PiperOrigin-RevId: 188651655
generated by cgit v1.2.3 (git 2.39.1) at 2024-07-09 04:53:25 +0000