-
Notifications
You must be signed in to change notification settings - Fork 74k
Insights: tensorflow/tensorflow
Overview
Could not load contribution data
Please try again later
246 Pull requests merged by 4 people
-
Integrate LLVM at llvm/llvm-project@e83adfe59632
#69818 merged
Jun 15, 2024 -
Reverts f4a4dbfa1a3c867ea7657a239b547829feb7157e
#69828 merged
Jun 15, 2024 -
Adds an optional lambda argument to the HloConstantSplitter pass. This allows
#69620 merged
Jun 15, 2024 -
Add tests for mlir_to_hlo `xla::Serialize` method. Fix bug for programs emitting CHLO.
#69751 merged
Jun 15, 2024 -
[XLA] Add shardings for implicit operands and return values of CaseOp and IfOp.
#69773 merged
Jun 15, 2024 -
[xla:gpu] Move the code that add SPMD pipeline to a utility function.
#69558 merged
Jun 15, 2024 -
Automated Code Change
#69824 merged
Jun 15, 2024 -
[xla:cpu] Check host kernel buffer arguments alignment
#69728 merged
Jun 15, 2024 -
Automated Code Change
#68741 merged
Jun 15, 2024 -
[xla:cpu] Add optimizer micro-benchmark
#69746 merged
Jun 15, 2024 -
[xla:cpu] Add a flag to set preferred vector width for LLVM backend
#69747 merged
Jun 15, 2024 -
[XLA:GPU] Extend `BlockLevelParameters` and simplify API to Triton IR emitter.
#69724 merged
Jun 15, 2024 -
[xla:cpu] NFC: Remove MLIRContext from dot emitter
#69755 merged
Jun 15, 2024 -
Add patch ahead of LLVM integrate to fix CI
#69805 merged
Jun 15, 2024 -
Pass GpuCompatibilityFlags to CheckGpuDelegateCompatibility.
#69809 merged
Jun 15, 2024 -
Update parser to map the inputs axis with the dynamic update slice kernel inputs.
#69569 merged
Jun 14, 2024 -
Remove incorrect comment.
#69744 merged
Jun 14, 2024 -
Add a tag to remove a few targets from internal code coverage computation.
#69787 merged
Jun 14, 2024 -
PR #13787: [GPU] Fix and cleanup cuDNN GEMM fusion tests.
#69781 merged
Jun 14, 2024 -
PR #13497: Swap inner and outer minor reduced dimension of tree reduction
#69774 merged
Jun 14, 2024 -
Integrate LLVM at llvm/llvm-project@da249cad8d39
#69766 merged
Jun 14, 2024 -
[XLA:GPU][MLIR-based indexing] Clean-up before removing tiling.
#69772 merged
Jun 14, 2024 -
Make sure that the same serialization is used for backend config.
#69763 merged
Jun 14, 2024 -
PR #13513: Prevent XLA crash in case if PATH variable is not set
#69761 merged
Jun 14, 2024 -
[XLA:GPU] Make Interval & IndexingMap properly hashable
#69764 merged
Jun 14, 2024 -
[XLA:GPU] Enable H100 for triton legacy support test
#69765 merged
Jun 14, 2024 -
PR #13687: Fix simplify_fp_conversions_test on Hopper
#69705 merged
Jun 14, 2024 -
[XLA:GPU][NFC] Remove irrelevant attributes in backend config of two tests.
#69696 merged
Jun 14, 2024 -
[xla:cpu] Add `ElementTypesSameAndSupported` helper function to `ThunkEmitter`
#69603 merged
Jun 14, 2024 -
PR #13760: Increase alignment of Traits::Params to 128
#69759 merged
Jun 14, 2024 -
[XLA:GPU] Dissociate the logic calling legacy Triton emitters from the one calling the new ones.
#69695 merged
Jun 14, 2024 -
[XLA:GPU][NFC] Cleanup logging for PGLE latency estimator.
#69698 merged
Jun 14, 2024 -
PR #13768: [XLA:GPU] Add synchronized allocation mode for cuda async memory allocator
#69749 merged
Jun 14, 2024 -
Rewrite the core logic of the computation partitioner.
#69693 merged
Jun 14, 2024 -
Integrate LLVM at llvm/llvm-project@46080abe9b13
#69731 merged
Jun 14, 2024 -
Eliminate StreamExecutor::RecordEvent by placing the required logic in Stream's derived classes.
#69647 merged
Jun 14, 2024 -
Integrate StableHLO at openxla/stablehlo@dd48ec58
#69715 merged
Jun 14, 2024 -
[XLA] HLOEvaluator tracing features
#69644 merged
Jun 14, 2024 -
[XLA:GPU] Clang-tidy cleanup for xla/service/all_reduce_folder.cc
#69655 merged
Jun 14, 2024 -
Improve the perf of tf.sparse.segment_(mean/sum/sqrt) ops when using
#69501 merged
Jun 14, 2024 -
Adding testing infrastructure for gather fusions.
#69741 merged
Jun 14, 2024 -
Update version numbers for TensorFlow 2.16.2
#69484 merged
Jun 14, 2024 -
[XLA:GPU] Clang-tidy cleanup for xla/service/all_gather_decomposer.h
#69652 merged
Jun 14, 2024 -
Remove `third_party/tf_runtime` from TSL
#69613 merged
Jun 14, 2024 -
Use xla::Shape to declare dynamism of arguments in tpu compile.
#69729 merged
Jun 13, 2024 -
[XLA:GPU] Extract Triton support test parsing and lookup boilerplate into a function.
#69683 merged
Jun 13, 2024 -
[XLA:GPU] Clang-tidy cleanup for xla/service/all_reduce_reassociate.cc
#69658 merged
Jun 13, 2024 -
[xla:cpu] Make CopyThunk parallel and asynchronous
#69663 merged
Jun 13, 2024 -
Update v2/setup.py
#69740 merged
Jun 13, 2024 -
Run `nvidia-smi` on GPU builds
#69735 merged
Jun 13, 2024 -
Improve core selector
#69500 merged
Jun 13, 2024 -
[PJRT] Use SerializeUsingVersionedStablehlo for PJRT API v47+
#69725 merged
Jun 13, 2024 -
Make `pxla.shard_arg` batch calls to `xc.copy_array_to_devices_with_sharding`
#69389 merged
Jun 13, 2024 -
Fix usage of `gunit_for_library_testonly` in service/gpu
#69721 merged
Jun 13, 2024 -
[XLA:GPU] Clang-tidy cleanup for xla/service/all_reduce_key.h
#69656 merged
Jun 13, 2024 -
Add Profiler::BlockedQueue Iterator *() and ->() operator const modifier
#69441 merged
Jun 13, 2024 -
[XLA:GPU] Clang-tidy cleanup for xla/service/all_gather_decomposer_test.cc
#69654 merged
Jun 13, 2024 -
Add broadcasting support
#69457 merged
Jun 13, 2024 -
Add TF op and name scope lines to derived timeline.
#69712 merged
Jun 13, 2024 -
[XLA] Remove LOG(INFO) prints from HostOffloader.
#69640 merged
Jun 13, 2024 -
Fix time_range filtering in trace viewer when only start_time is specified.
#69722 merged
Jun 13, 2024 -
Don't use RBE for continuous builds
#69717 merged
Jun 13, 2024 -
[xla:cpu] Add a flag to enable concurrency-optimized scheduler
#69662 merged
Jun 13, 2024 -
Internal BUILD file changes.
#69559 merged
Jun 13, 2024 -
Add a test that verifies that the arrays returned from `Client::CopyArrays` has correct sharding objects
#69713 merged
Jun 13, 2024 -
r2.16 cherry-pick: 264dce9fb38 "Remove CMake from requirements now that dm-tree has 3.12 wheels."
#69726 merged
Jun 13, 2024 -
Add is_ici_weight_dist attribute to ops created in xla_broadcast pass.
#69638 merged
Jun 13, 2024 -
Hook up XNNPack per channel quantized deconvolution
#69692 merged
Jun 13, 2024 -
[xla:cpu] Add benchmark for bcast + multiply
#69714 merged
Jun 13, 2024 -
[XLA:GPU][NFC] Re-enable BF16 tests post-Ampere.
#69710 merged
Jun 13, 2024 -
[XLA:Python] Use a higher stacklevel in pytree deprecation warning.
#69709 merged
Jun 13, 2024 -
Rename hlo_module_id to program_id in XLA Op TraceMe
#69648 merged
Jun 13, 2024 -
[XLA:GPU] Clang-tidy cleanup for xla/service/all_reduce_folder_test.cc
#69657 merged
Jun 13, 2024 -
Add flags to gpu_compatibility
#69630 merged
Jun 13, 2024 -
[XLA:GPU][MLIR-based indexing] Move constructors' common parts to MlirReductionFusion().
#69685 merged
Jun 13, 2024 -
[xla:cpu] Run transitive reduction to remove redundant edges from ThunkExecutor
#69664 merged
Jun 13, 2024 -
Integrate LLVM at llvm/llvm-project@98174fb6ec97
#69686 merged
Jun 13, 2024 -
[XLA:GPU] Allow Triton emitter to handle fp8 matmuls.
#69531 merged
Jun 13, 2024 -
Register TPU costs in the new batch_stats module.
#66962 merged
Jun 13, 2024 -
Generate XNNPack cache flatbuffers header in the CMake build.
#69594 merged
Jun 13, 2024 -
[XLA:GPU][NFC] Use fusion_root() instead of fusion_roots()[].
#69621 merged
Jun 13, 2024 -
[xla:cpu] Pass ThunkEmitter's constructor parameters by reference instead of pointers.
#69690 merged
Jun 13, 2024 -
[XLA:CPU] Remove unused proto
#69608 merged
Jun 13, 2024 -
[XLA:GPU][NFC] Move GPU specific latency hiding scheduler components to a separate file.
#69679 merged
Jun 13, 2024 -
[xla:cpu] Pass HLO module config and target machine features to `ThunkEmitter`
#69687 merged
Jun 13, 2024 -
[XLA:GPU] Simplify Triton Support test preamble and BF16 skip logic
#69682 merged
Jun 13, 2024 -
Introduce the batch_stats module.
#68601 merged
Jun 13, 2024 -
[NFC] Explicitly set the value of --xla_gpu_mlir_emitter_level.
#69684 merged
Jun 13, 2024 -
Make TestOddElements pass with MLIR emitters.
#69677 merged
Jun 13, 2024 -
PR #13278: Fix _xla_send_recv_validation_attribute in loop-double-buffer-transformer
#69609 merged
Jun 13, 2024 -
PR #13646: [ROCM] gemm_rewriter: bugfixing supported datatypes combinations
#69676 merged
Jun 13, 2024 -
PR #13411: [XLA:GPU][Allow cuda async allocator to use non-default pool
#69675 merged
Jun 13, 2024 -
Explicitly set the value of --xla_gpu_mlir_emitter_level.
#69673 merged
Jun 13, 2024 -
[XLA:GPU][NFC] Remove std::functions for shmem indices.
#69611 merged
Jun 13, 2024 -
[TFQ] Make runtime_client_py visible to all sub packages of tfq
#69641 merged
Jun 13, 2024 -
[tsl:concurrency] NFC: Remove KeepAsyncValuePayloadOnError alias
#69666 merged
Jun 13, 2024 -
Give `TF_CUDA_COMPUTE_CAPABILITY` via `repo_env` instead of `action_env`
#69639 merged
Jun 13, 2024 -
expose "bad_indice_policy" attribute for tf.gather_nd (python API)
#68802 merged
Jun 13, 2024 -
Use intersect to combine TPU step events in OSS
#69204 merged
Jun 13, 2024 -
Introduce `Client::CopyArrays()` for batched device-to-device copy
#69096 merged
Jun 13, 2024 -
[xla:cpu] Use ThunkExecutor to execute nested thunk sequences inside control flow thunks
#69631 merged
Jun 13, 2024 -
[XLA:GPU] Clang-tidy cleanup for xla/service/gpu/cudnn_fusion_compiler.h
#69649 merged
Jun 13, 2024 -
Automated Code Change
#69605 merged
Jun 13, 2024 -
Add exhaustive Reciprocal numerics tests
#69636 merged
Jun 13, 2024 -
Update TFRT dependency to use revision
#69637 merged
Jun 13, 2024 -
[xla:cpu] Make ThunkExecutor::Execute asynchronous
#69604 merged
Jun 12, 2024 -
Split error_reporter dependency from tflite compiler quantization
#69194 merged
Jun 12, 2024 -
Fix typo in `quantum`
#69642 merged
Jun 12, 2024 -
Add XLA:CPU module lines to the host trace.
#69566 merged
Jun 12, 2024 -
Update version in setup.py to 2.16.2
#69643 merged
Jun 12, 2024 -
Remove the usage of DynamicBuffer from flatbuffer_export.cc
#69623 merged
Jun 12, 2024 -
Add 'decompose_optionals' pass.
#69563 merged
Jun 12, 2024 -
PR #13662: [ROCm] Fix build break of cudnn_fused_conv_rewriter_test due to `1268712`
#69610 merged
Jun 12, 2024 -
Use absl::StatusOr rather than the xla::StatusOr alias, since they're now identical.
#69420 merged
Jun 12, 2024 -
[XLA:GPU] Use predication instead of branching in topk_kernel.
#69614 merged
Jun 12, 2024 -
Remove prefetch(0) from noop_elimination.
#69625 merged
Jun 12, 2024 -
Use tensorboard nightly 2.18.0a
#69562 merged
Jun 12, 2024 -
Reverts 5969602d6a1114d538aa98dbde9f27304fc6b22f
#69619 merged
Jun 12, 2024 -
[XLA] Allow for turning off fast add path for reduce
#69616 merged
Jun 12, 2024 -
[xla:cpu] Pass thread pool to thunk execution.
#69589 merged
Jun 12, 2024 -
Update TFRT dependency to use revision
#69250 merged
Jun 12, 2024 -
Fix gradient path for tf.sparse.segment_{sum/mean/sqrt) and bfloat16/float16
#69502 merged
Jun 12, 2024 -
Update lock files
#69615 merged
Jun 12, 2024 -
TraceMe: Use std::is_invocable_v instead of custom implementation
#69537 merged
Jun 12, 2024 -
r2.17 cherry-pick: 8fde1f4381a "Update TensorFlow ml_dtypes dependency to >= 0.3.1 < 0.5.0"
#69612 merged
Jun 12, 2024 -
[XLA:FFI] Move stream insertion operator for C API datatypes to global namespace.
#68518 merged
Jun 12, 2024 -
[XLA:GPU][NFC] Add a `BlockLevelParameters` type and a `BlockLevelFusionConfig` proto message.
#69601 merged
Jun 12, 2024 -
[tsl:concurrency] NFC: Remove KeepAsyncValuePayloadOnError alias
#69554 merged
Jun 12, 2024 -
PR #13569: [GPU] Add on-disk per-kernel compilation cache.
#69597 merged
Jun 12, 2024 -
Enable hlo_module_config_test for OSS testing
#69588 merged
Jun 12, 2024 -
[XLA:GPU] Extract legacy dot and dynamic_slice tests out of the triton_support_test.
#69593 merged
Jun 12, 2024 -
[XLA] [NFC] Remove unused Client APIs
#69584 merged
Jun 12, 2024 -
Integrate LLVM at llvm/llvm-project@c012e487b724
#69583 merged
Jun 12, 2024 -
[XLA:GPU] Simplify triton support tests - remove unnecessary SoftMax
#69587 merged
Jun 12, 2024 -
Move common test helper to test_utils.h
#69578 merged
Jun 12, 2024 -
Delay using tuple arguments in Shardonnay until after it's run.
#69477 merged
Jun 12, 2024 -
Automated Code Change
#69522 merged
Jun 12, 2024 -
Fix fallback to RunHloModuleIterationLiterals.
#69527 merged
Jun 12, 2024 -
[XLA:GPU] Better approximation for costly AR in LHS.
#69535 merged
Jun 12, 2024 -
PR #13639: Fix UAF in Norm Rewriter
#69579 merged
Jun 12, 2024 -
Automated Code Change
#69444 merged
Jun 12, 2024 -
PR #13477: Add SetupDerivedInstruction for whileop in spmd partitioner
#69539 merged
Jun 12, 2024 -
PR #11583: Weight offloading of Jax memories: support memory_kind for GPUs
#69576 merged
Jun 12, 2024 -
Automated Code Change
#69421 merged
Jun 12, 2024 -
[tsl:concurrency] Update AsyncValueRef documentation and add more implicit constructors
#69570 merged
Jun 12, 2024 -
Automated Code Change
#69446 merged
Jun 12, 2024 -
[xla:cpu] Add benchmark for dag execution
#69549 merged
Jun 12, 2024 -
[xla:cpu] Add proper error handling to ThunkExecutor
#69556 merged
Jun 12, 2024 -
Reverts c2e7e9f6c3f4d4937d8145f988ea74818e000ecc
#69571 merged
Jun 12, 2024 -
[xla:cpu] Add error handling to async host kernel launch
#69555 merged
Jun 12, 2024 -
1. Simplify a redundant code snippet
#69565 merged
Jun 12, 2024 -
Reverts changelist 641306427
#69567 merged
Jun 12, 2024 -
Move JAX builds to build.py
#69542 merged
Jun 11, 2024 -
[tf.data] Add `synchronous` parameter to `map`.
#69543 merged
Jun 11, 2024 -
[XLA] remove degenerate indexing dimensions in algebraic simplifier
#69538 merged
Jun 11, 2024 -
[xla:cpu] Return thunk completion as an async event
#69489 merged
Jun 11, 2024 -
Bump minSdkVersion of TFL Android libs to 21.
#69560 merged
Jun 11, 2024 -
#tf-data Turn up `map_fusion` experiment to 50% task-level.
#69548 merged
Jun 11, 2024 -
Remove unused options argument from Platform::Initialize.
#69498 merged
Jun 11, 2024 -
Move XplaneEventMutator and XplaneEventMutatorFactory to a separate file.
#69547 merged
Jun 11, 2024 -
Integrate LLVM at llvm/llvm-project@8c5d9c79b96e
#69530 merged
Jun 11, 2024 -
Add GetLevelForDuration() helper function.
#69410 merged
Jun 11, 2024 -
[tsl:concurrency] Add optional implicit conversion from absl::Status to ErrorAsyncValue
#69490 merged
Jun 11, 2024 -
[XLA] [NFC] Remove redundant Status from ToProtoWithConfig
#69462 merged
Jun 11, 2024 -
Update TensorFlow ml_dtypes dependency to >= 0.3.1 < 0.5.0
#69528 merged
Jun 11, 2024 -
[XLA:GPU][NFC] Add tests for AllReduceSplitter with a following GpuReduceScatterCreator.
#69474 merged
Jun 11, 2024 -
Migrate ConvertMlirToGraphdef uses to tf2xla version and deprecate translate version.
#69544 merged
Jun 11, 2024 -
[XLA] Add layout constraint custom call and simplify it away after layout
#69495 merged
Jun 11, 2024 -
[XLA:GPU] Initial simplification of Triton Support test.
#69534 merged
Jun 11, 2024 -
Fix Release notes
#69540 merged
Jun 11, 2024 -
[XLA:GPU] Extract useful utils from ir_emitter_triton_test into their own file.
#69533 merged
Jun 11, 2024 -
Add bazelrc change that should've gone in with https://github.com/openxla/xla/pull/13408
#69509 merged
Jun 11, 2024 -
Fix xprofilez integration_tests:xprofilez_handler_gpu_test fail
#69511 merged
Jun 11, 2024 -
[XLA] Fix an incorrect use of a hashmap in HloCSE.
#69526 merged
Jun 11, 2024 -
Clean up sparsity patches.
#69525 merged
Jun 11, 2024 -
[XLA] [NFC] Remove unused Status
#69383 merged
Jun 11, 2024 -
Automated Code Change
#69515 merged
Jun 11, 2024 -
Add support for input_literals_file and output_literals_file options.
#69524 merged
Jun 11, 2024 -
PR #13108: [GPU] Shard GEMM fusion autotuning across multiple compilation processes.
#69521 merged
Jun 11, 2024 -
PR #13502: [ROCm] Fix collective ops test
#69518 merged
Jun 11, 2024 -
Automated Code Change
#69520 merged
Jun 11, 2024 -
Update zlib to 1.3.1
#69519 merged
Jun 11, 2024 -
Actually support the flag xla_dump_large_constants.
#69513 merged
Jun 11, 2024 -
PR #13568: [GPU] Let constants be emitted into a separate LLVM module.
#69514 merged
Jun 11, 2024 -
Automated Code Change
#69438 merged
Jun 11, 2024 -
Automated Code Change
#69505 merged
Jun 11, 2024 -
Automated Code Change
#69506 merged
Jun 11, 2024 -
Swap operands of dot if the LHS is fed by a parameter
#68768 merged
Jun 11, 2024 -
Add new stat to MegaScaleStatTypeMap.
#69402 merged
Jun 11, 2024 -
Remove unused PlatformManagerImpl::InitializePlatformWithName method.
#69494 merged
Jun 11, 2024 -
Bump API version
#69400 merged
Jun 10, 2024 -
Add support for int4 in dequantize op.
#68572 merged
Jun 10, 2024 -
Fix compute capability as given in `TF_CUDA_COMPUTE_CAPABILITIES`
#69493 merged
Jun 10, 2024 -
Set `build.repo` like "openxla/xla` rather than just "xla"
#69491 merged
Jun 10, 2024 -
[XLA:DataFlowAnalysis] Add unit tests for loop/output fusions with a dynamic-update-slice and collective.
#69488 merged
Jun 10, 2024 -
Preserve HloModuleConfig in HLO<->MHLO.
#69349 merged
Jun 10, 2024 -
Set compute_capability via action_env correctly
#69486 merged
Jun 10, 2024 -
SparsifyModel returns absl::Status instead of TfLiteStatus.
#68902 merged
Jun 10, 2024 -
PR #12892: Add collective-permute-valid-iteration-annotator
#69109 merged
Jun 10, 2024 -
Refactor in preparation for moving JAX/XLA CI to build.py
#69405 merged
Jun 10, 2024 -
[HLO] Use .hlo for HLO text files, remove uses of .hlotxt.
#69262 merged
Jun 10, 2024 -
XProf GPU: Using Per-Thread callback api data for CUPTI collector for better overhead.
#65009 merged
Jun 10, 2024 -
Add no_gl library equivalents for gpu_api_delegate
#69479 merged
Jun 10, 2024 -
Allow generation of sharding strategies with mixed mesh shapes by default.
#69401 merged
Jun 10, 2024 -
Update release notes for TensorFlow 2.16.2
#69396 merged
Jun 10, 2024 -
Added an API call for registering external types with XLA:FFI
#69465 merged
Jun 10, 2024 -
[xla:cpu] NFC: Add factory constructor to all CPU thunks
#69473 merged
Jun 10, 2024 -
[XLA:GPU][NFC] Replace `ABSL_ATTRIBUTE_UNUSED` with `[[maybe_unused]]`.
#69476 merged
Jun 10, 2024 -
[xla:cpu] Add more tests and benchmarks for ThunkExecutor and fix tsan races
#69450 merged
Jun 10, 2024 -
[PJRT][IFRT] Move topology discovery into PJRT-IFRT.
#68260 merged
Jun 10, 2024 -
Save function argument names as locs through MHLO<->HLO conversion.
#68605 merged
Jun 10, 2024 -
Fix simplification of modulo with a negative multiplier on the LHS.
#69467 merged
Jun 10, 2024 -
Remove the standalone python autotuner for Triton
#69468 merged
Jun 10, 2024 -
Combine sparsity patches after integration
#69466 merged
Jun 10, 2024 -
PR #13372: [GPU] Fix autotuner_util_test.
#69380 merged
Jun 10, 2024 -
Replace Interval::NumElements with ::GetLoopTripCount.
#69463 merged
Jun 10, 2024 -
[XLA:GPU] Match more sort cases in the GPU Sort Rewriter.
#69412 merged
Jun 10, 2024 -
PR #13542: [NVIDIA] Change DCE to replay control deps for GTE-fusion simplification.
#69458 merged
Jun 10, 2024 -
Don't attempt to vectorize complex reductions.
#69460 merged
Jun 10, 2024 -
Stop using xla::Status alias for absl::Status. Just use absl::Status instead.
#69419 merged
Jun 10, 2024 -
Automated Code Change
#69442 merged
Jun 10, 2024 -
[xla:cpu] Add completion event to Thunk and prepare for making them async
#69411 merged
Jun 10, 2024 -
[xla:cpu] Add fusion benchmark
#69409 merged
Jun 10, 2024 -
[xla:cpu] Use recursive work splitting to submit host tasks
#69403 merged
Jun 10, 2024 -
Improvement of the regular expression
#62309 merged
Jun 10, 2024 -
[xla:cpu] Add async Launch to HostKernel and use Eigen device to parallelize kernel execution
#69348 merged
Jun 10, 2024 -
[xla:cpu] Emit partitioned loops if operation marked for parallel execution
#69452 merged
Jun 9, 2024 -
Automated Code Change
#69436 merged
Jun 9, 2024
130 Pull requests opened by 5 people
-
Automated Code Change
#69439 opened
Jun 9, 2024 -
Automated Code Change
#69440 opened
Jun 9, 2024 -
Automated Code Change
#69445 opened
Jun 9, 2024 -
Automated Code Change
#69447 opened
Jun 9, 2024 -
Automated Code Change
#69449 opened
Jun 9, 2024 -
Automated Code Change
#69451 opened
Jun 9, 2024 -
Batch `pxla.shard_args` calls triggered by `jax.device_put`
#69453 opened
Jun 9, 2024 -
Automated Code Change
#69454 opened
Jun 10, 2024 -
[tsl] platform/logging/default: do not capture LOG messages with VLogFileMgr
#69456 opened
Jun 10, 2024 -
Disable AMD test that broke TAP due to UBSAN issues
#69469 opened
Jun 10, 2024 -
[XLA:GPU] NFC - Remove a couple of functions from the API of triton_support.
#69472 opened
Jun 10, 2024 -
XProf GPU: Using Per-Thread callback api data for CUPTI collector for better overhead.
#69481 opened
Jun 10, 2024 -
Add an option to disable restoring variables when there is saver_def in SavedModel loader.
#69482 opened
Jun 10, 2024 -
Add more custom pattern to hlo_unstacker pass.
#69485 opened
Jun 10, 2024 -
SparsifyModel returns absl::Status instead of TfLiteStatus.
#69487 opened
Jun 10, 2024 -
dump: Add a flag to explicitly disable dumping.
#69496 opened
Jun 10, 2024 -
Integrate LLVM at llvm/llvm-project@bb2bf3a2635a
#69499 opened
Jun 11, 2024 -
Automated Code Change
#69507 opened
Jun 11, 2024 -
Add support for Pathways topology in GpuTopology.
#69508 opened
Jun 11, 2024 -
Automated Code Change
#69512 opened
Jun 11, 2024 -
PR #13486: [ROCm] Add AMD_SERIALIZE_KERNEL environment for gpu_offloading_test
#69517 opened
Jun 11, 2024 -
PR #13569: [GPU] Add on-disk per-kernel compilation cache.
#69523 opened
Jun 11, 2024 -
Turn on layer scanning for llama2-7b on GPU.
#69536 opened
Jun 11, 2024 -
Adopt ConvertMlirHloToHloModule instead of passing in proto in PJRT
#69551 opened
Jun 11, 2024 -
[Multi-host GPU] Build GpuTopology only by device ids when the topology is asymmetric.
#69552 opened
Jun 11, 2024 -
Integrate LLVM at llvm/llvm-project@3af35251c8cd
#69557 opened
Jun 11, 2024 -
Reverts 49f4d9052259ae562ad0b0b84b5ba759494e6f83
#69564 opened
Jun 11, 2024 -
Remove the deprecated PjRtClient::LookupAddressableDevice() that takes a raw int.
#69572 opened
Jun 12, 2024 -
Integrate LLVM at llvm/llvm-project@3af35251c8cd
#69573 opened
Jun 12, 2024 -
PR #13411: [XLA:GPU][Allow cuda async allocator to use non-default pool
#69574 opened
Jun 12, 2024 -
PR #13569: [GPU] Add on-disk per-kernel compilation cache.
#69577 opened
Jun 12, 2024 -
PR #13462: [ROCM][NFC] gpublas-lt refactoring after adding workspace and scratch allocator
#69580 opened
Jun 12, 2024 -
[XLA] Remove ServiceInterface
#69581 opened
Jun 12, 2024 -
Integrate LLVM at llvm/llvm-project@c012e487b724
#69582 opened
Jun 12, 2024 -
[XLA:GPU] Add initial SymbolicTileAnalysis::GetGoodTilings implementation
#69591 opened
Jun 12, 2024 -
PR #13467: [GPU] Relax the check for scheduled modules.
#69599 opened
Jun 12, 2024 -
[XLA] Remove unused Client API/proto msg
#69600 opened
Jun 12, 2024 -
PR #13310: [NVIDIA GPU] Added a rewrite logic in gpu_windowned_einsum_handler to handle all2all
#69606 opened
Jun 12, 2024 -
Fork jet_gpu_compatibility
#69617 opened
Jun 12, 2024 -
[XLA:GPU] Don't pass produce and consumer run time data to EstimateRunTimeForFusion.
#69622 opened
Jun 12, 2024 -
Only split those constants that are shared between manually and automatically sharded regions of the graph.
#69626 opened
Jun 12, 2024 -
[xla] add missing includes for absl::StrCat
#69627 opened
Jun 12, 2024 -
[XLA:GPU] Use predication instead of branching in topk_kernel.
#69628 opened
Jun 12, 2024 -
Remove dependency on runtime string_utils.
#69629 opened
Jun 12, 2024 -
Remove usage of tflite::ControlEdges from flatbuffer_export.cc
#69632 opened
Jun 12, 2024 -
Add configs for Tensorflow Kokoro builds on XLA
#69634 opened
Jun 12, 2024 -
Move TF dependence from TFL tflite_copts
#69646 opened
Jun 12, 2024 -
[XLA:GPU] Clang-tidy cleanup for xla/service/all_gather_broadcast_reorder.cc
#69653 opened
Jun 13, 2024 -
[XLA:GPU] Clang-tidy cleanup for xla/service/all_reduce_simplifier.cc
#69660 opened
Jun 13, 2024 -
expose "bad_indice_policy" attribute for tf.gather_nd (python API)
#69667 opened
Jun 13, 2024 -
[PJRT][Fix] Add device check for NextPluggableDevice in GetPjRtExecuteOptions
#69669 opened
Jun 13, 2024 -
Automated Code Change
#69670 opened
Jun 13, 2024 -
Updated API docs for tf.quantization.fake_quant_with_min_max_args_gradient with Example code.
#69671 opened
Jun 13, 2024 -
Use pthreadpool only when XNNPACK is enabled
#69672 opened
Jun 13, 2024 -
[XLA] Remove proto-based communication for service/client
#69688 opened
Jun 13, 2024 -
Reverts 8164fe4fd6ac35789d584e331f137a0fa1e173a2
#69691 opened
Jun 13, 2024 -
[XLA] Reverse the direction of dependency arrows for XLA protos
#69694 opened
Jun 13, 2024 -
[XLA:GPU] Improve error message for missing costs/latencies in PGLE.
#69697 opened
Jun 13, 2024 -
[xla:cpu] Add FFT thunk
#69703 opened
Jun 13, 2024 -
Integrate LLVM at llvm/llvm-project@846103c7e389
#69704 opened
Jun 13, 2024 -
Disable convolution algorithm 14
#69706 opened
Jun 13, 2024 -
Move metadata_util to utils folder.
#69708 opened
Jun 13, 2024 -
PR #13721: Fix BUILD files to allow building //xla/service/gpu/... in OSS
#69711 opened
Jun 13, 2024 -
Migrate usage of schema_conversion_utils.
#69716 opened
Jun 13, 2024 -
[XLA:GPU] Simplify TritonSupport tests by providing a standard ENTRY computation.
#69720 opened
Jun 13, 2024 -
[XLA] Remove LOG(INFO) prints from HostOffloader.
#69727 opened
Jun 13, 2024 -
[XLA:MSA] Sync copy replacement
#69730 opened
Jun 13, 2024 -
Integrate LLVM at llvm/llvm-project@846103c7e389
#69732 opened
Jun 13, 2024 -
When determining constant value we should use the constant values stored in the SetBound custom call.
#69733 opened
Jun 13, 2024 -
Upstream flatbuffer utils to read big models
#69736 opened
Jun 13, 2024 -
[PJRT] Use SerializeUsingVersionedStablehlo for PJRT API v47+
#69737 opened
Jun 13, 2024 -
Update quantizer inputs to read in bytearray
#69738 opened
Jun 13, 2024 -
Created a new LegalizeMlirToHloReproducer proto and dump that instead of just the mlir module.
#69742 opened
Jun 13, 2024 -
PR #13482: [ROCm] Distinguish between NVIDIA and AMD gpu tags
#69743 opened
Jun 13, 2024 -
[IFRT] Add AttributeMap
#69745 opened
Jun 14, 2024 -
Fix a bug in xplane_to_step_events.cc.
#69748 opened
Jun 14, 2024 -
Increase default safety of Keras tar extraction
#69752 opened
Jun 14, 2024 -
Updated tensorflow/tensorflow/lite/python/lite.py with grammatical a…
#69756 opened
Jun 14, 2024 -
Integrate LLVM at llvm/llvm-project@46080abe9b13
#69760 opened
Jun 14, 2024 -
PR #13603: NVTX: name threads, CUDA devices and CUDA streams
#69762 opened
Jun 14, 2024 -
PR #13603: NVTX: name threads, CUDA devices and CUDA streams
#69767 opened
Jun 14, 2024 -
[XLA:GPU] Keep hashmap consistent in RescaleSymbols
#69768 opened
Jun 14, 2024 -
Refactor and Enhance TensorFlow Function Execution Code
#69769 opened
Jun 14, 2024 -
Add New Features and Enhance TensorFlow ConstOp Tests
#69770 opened
Jun 14, 2024 -
Move BlockedSparseToMMA pattern from Triton to XLA.
#69771 opened
Jun 14, 2024 -
[XLA:GPU] Enable new mlir loop emitter by default.
#69775 opened
Jun 14, 2024 -
Introduce nested tuple support in FFI
#69776 opened
Jun 14, 2024 -
[XLA:GPU] Make BufferComparator accept tolerance as a parameter
#69780 opened
Jun 14, 2024 -
Add test case for 1D convolution
#69782 opened
Jun 14, 2024 -
[TSL] Remove apparently unnecessary "template" keywords that are yielding a clang warning.
#69783 opened
Jun 14, 2024 -
PR #13722: [ROCM] rocBLAS: default algorithm fallback
#69784 opened
Jun 14, 2024 -
Remove unused includes from `context.h` and `context_test.cc`.
#69785 opened
Jun 14, 2024 -
[XLA:GPU] Add initial version of cost model for tiled hlo.
#69788 opened
Jun 14, 2024 -
Integrate LLVM at llvm/llvm-project@e83adfe59632
#69791 opened
Jun 14, 2024 -
Modify boot_id per LocalTopology when using mock NCCL
#69792 opened
Jun 14, 2024 -
Go back to old continuous build until L4 RBE is ready
#69793 opened
Jun 14, 2024 -
Reverts 27125ab80d84e1a9f0e0d93aa5416e316c73e91d
#69794 opened
Jun 14, 2024 -
Throwaway change for presubmit tests
#69795 opened
Jun 14, 2024 -
PR #13190: Add pipelined while loop annotator
#69797 opened
Jun 14, 2024 -
Reverts changelist 578813627
#69798 opened
Jun 14, 2024 -
Bump XLA Docker Python version 3.11 -> 3.12
#69801 opened
Jun 14, 2024 -
#tf-data-service Improve alternative data transfer API in worker config.
#69802 opened
Jun 14, 2024 -
Revert "Move JAX builds to build.py"
#69803 opened
Jun 14, 2024 -
Integrate LLVM at llvm/llvm-project@9b7b1bee07ea
#69806 opened
Jun 14, 2024 -
Add `AsyncWrapper` pass to the `GpuCompiler` to wrap `dot` operations.
#69807 opened
Jun 14, 2024 -
Introduce AsyncWrapper.
#69808 opened
Jun 14, 2024 -
[XLA] Allow propagations through broadcasts
#69810 opened
Jun 15, 2024 -
[tf] ortools/scip: update diff to v8.0.3 baseline
#69813 opened
Jun 15, 2024 -
Reverts dd6e541267d0ce9d4f80216f9b9e91f404939124
#69814 opened
Jun 15, 2024 -
Remove `mhlo_quant_legalize_to_int pass` Pass from openxla/mhlo
#69815 opened
Jun 15, 2024 -
Stop using xla/statusor.h now that it just contains an alias for absl::Status.
#69816 opened
Jun 15, 2024 -
Automated Code Change
#69817 opened
Jun 15, 2024 -
Stop using xla/statusor.h now that it just contains an alias for absl::Status.
#69819 opened
Jun 15, 2024 -
Stop using xla/statusor.h now that it just contains an alias for absl::Status.
#69821 opened
Jun 15, 2024 -
[xla:cpu] Add support for lowering dot operations to kernel or dot thunk
#69822 opened
Jun 15, 2024 -
[XLA] Add shardings for implicit operands and return values of CaseOp and IfOp.
#69826 opened
Jun 15, 2024 -
Implements the `Reshard` method for the BasicStringArray class.
#69829 opened
Jun 15, 2024 -
[xla:cpu] Add dot benchmark and enable ThreadPoolDevice and contraction kernel in DotThunk
#69830 opened
Jun 15, 2024 -
Automated Code Change
#69831 opened
Jun 15, 2024 -
[xla:cpu] Split DotThunk to enable parallel compilation
#69832 opened
Jun 15, 2024 -
[XLA] Make HLO instrumentation respect execution threads
#69833 opened
Jun 15, 2024
30 Issues closed by 12 people
-
Improve tf_compile to allow emitting HLO module instead of/in addition to executable
#26627 closed
Jun 15, 2024 -
tf.train.BytesList should accept bytearray as an input type
#27047 closed
Jun 15, 2024 -
tensorflow[and-cuda] 2.15.0/2.15.1 compatibility with jax[cuda12]
#68290 closed
Jun 15, 2024 -
Trouble Running TensorFlow v2.16.1 with NVIDIA GeForce 940MX GPU #914
#68696 closed
Jun 15, 2024 -
Wrong explanation about an argument of tflite interpreter
#68862 closed
Jun 15, 2024 -
-
#69796 closed
Jun 14, 2024 -
Build from source C doesn't produce .tar.gz archive
#69266 closed
Jun 14, 2024 -
module 'keras.src.backend' has no attribute 'convert_to_numpy'
#66966 closed
Jun 14, 2024 -
Can not find strtod_l function on Android device
#61951 closed
Jun 13, 2024 -
Documentation for `tf.dynamic_partition()` has examples without a call to the API being described
#68259 closed
Jun 13, 2024 -
tf.nn.embedding_lookup works fine in CPU mode, but lacks constraint checking in GPU mode
#62628 closed
Jun 12, 2024 -
Tensorflow not detecting GPU
#64881 closed
Jun 12, 2024 -
stuck at installing tensorflow
#67067 closed
Jun 12, 2024 -
YOLOv5n model does work on python tflite but not C++ tflite
#69510 closed
Jun 12, 2024 -
How to reduce the time running invoke()
#68424 closed
Jun 12, 2024 -
Tensorflow Developer certificate didnt recieved yet
#68654 closed
Jun 12, 2024 -
"ImportError: random_device could not be read" when importing duckdb after importing tensorflow
#61741 closed
Jun 11, 2024 -
[feature] Smarter Handling of Image Data Format
#8227 closed
Jun 11, 2024 -
No gradient defined for operation 'MatrixExponential' (op type: MatrixExponential)
#15465 closed
Jun 11, 2024 -
embed_sequence and embedding_lookup behave differently on CPU vs. GPU
#17417 closed
Jun 11, 2024 -
Unable to install old version of tensorflow
#66950 closed
Jun 11, 2024 -
Aborted (core dumped) in `tf.raw_ops.IRFFTND\RFFTND\FFTND\IFFTND`
#68648 closed
Jun 10, 2024 -
.
#69022 closed
Jun 9, 2024 -
Multi-arch docker images
#14934 closed
Jun 9, 2024 -
Can't use CTCBeamSearchDecoder in c++, LINK ERROR occur,BUG in CTCBeamSearchDecoder 's source code
#22894 closed
Jun 9, 2024 -
Intermittent very long latency in XRT operations
#22975 closed
Jun 9, 2024 -
axis argument for FFT ops (tf.signal.fft, tf.signal.fft2d, etc.)
#23156 closed
Jun 9, 2024 -
IllegalArgumentException: Internal error: Failed to run on the given Interpreter
#66594 closed
Jun 9, 2024 -
aarch64/arm64 Tensorflow Lite runtime wheels not compatible with silicon OSX ?
#67422 closed
Jun 9, 2024
40 Issues opened by 30 people
-
Non-deprecated tf.keras.preprocessing alternatives don't cover properly all the deprecated features
#69834 opened
Jun 15, 2024 -
Cannot take the length of shape with unknown rank.
#69827 opened
Jun 15, 2024 -
Too many duplicate debug logs
#69825 opened
Jun 15, 2024 -
In tflite, how to use the same memory to serve different models with exactly the same structure
#69823 opened
Jun 15, 2024 -
Help Needed: AttributeError in tf2onnx Conversion from ONNX to TensorFlow Model
#69812 opened
Jun 15, 2024 -
Immediate Assistance Required: Issue with Converting Keras Model to TFLite
#69811 opened
Jun 15, 2024 -
Update curl from 8.4.0 to 8.6.0 due to security vulnerabilities CVE-2023-46219 and CVE-2023-46218
#69799 opened
Jun 14, 2024 -
[Feature Request] Batch Renormalization
#69790 opened
Jun 14, 2024 -
GCS gfile operations fail in TF nightly 2.17 and 2.18 when not running in GCP
#69789 opened
Jun 14, 2024 -
error: no such file or directory: 'v2' when Bazel build for macOS
#69786 opened
Jun 14, 2024 -
Aborted (core dumped) in `tf.raw_ops.CropAndResizeGradImage`
#69778 opened
Jun 14, 2024 -
tf.py_function does not output ragged tensors
#69777 opened
Jun 14, 2024 -
CMake Error: could not find requested file BuildFlatBuffers when cmake the lite kernel test
#69754 opened
Jun 14, 2024 -
CMake Error: could not find requested file BuildFlatBuffers when cmake the lite kernel test
#69753 opened
Jun 14, 2024 -
What is preventing TF to use GPU when used in native windows?
#69750 opened
Jun 14, 2024 -
Object Detection in Android using front camera: the detected bounding boxes are drawn incorrectly
#69734 opened
Jun 13, 2024 -
Rescaling Layer Issue when Loading .keras Model
#69719 opened
Jun 13, 2024 -
Installing Tensorflow on Fedora 40
#69718 opened
Jun 13, 2024 -
Significant Performance Drop When Training Sequential Model Using `tf.data.Dataset.from_generator`
#69702 opened
Jun 13, 2024 -
Aborted (core dumped) in `tf.raw_ops.BatchFunction`
#69701 opened
Jun 13, 2024 -
Segmentation fault in `tf.raw_ops.CollectiveAllToAllV2`
#69700 opened
Jun 13, 2024 -
Segmentation fault in `tf.raw_ops.CollectiveGatherV2`
#69699 opened
Jun 13, 2024 -
Cannot cross-compile minimal tflite example
#69689 opened
Jun 13, 2024 -
TFlite usage on android
#69680 opened
Jun 13, 2024 -
Fixed error in code on official TF guide for Transfer Learning and Fine Tuning
#69678 opened
Jun 13, 2024 -
"invalid static_cast" on AVX512FP16 (e.g. Sapphire Rapids)
#69674 opened
Jun 13, 2024 -
why tf use clang as the default compiler on windows?
#69665 opened
Jun 13, 2024 -
tf.signal.rfftnd throws NotFoundError on CPU execution (GPU-behavior unknown)
#69595 opened
Jun 12, 2024 -
tf.data.Dataset.save and load custom dataset throw "DataLossError: Unable to parse tensor from stored proto"
#69575 opened
Jun 12, 2024 -
SparseTensor Batching Utility
#69541 opened
Jun 11, 2024 -
Selectively Build TensorFlow Lite with Docker ends with an error.
#69532 opened
Jun 11, 2024 -
Transfer learning and fine-tuning doc seems to have unexpected results
#69480 opened
Jun 10, 2024 -
Aborted (core dumped) in `tf.raw_ops.SparseReduceSum\tf.raw_ops.SparseReduceMax`
#69470 opened
Jun 10, 2024 -
16KB so support
#69459 opened
Jun 10, 2024 -
Crash in `tf.raw_ops.SparseCountSparseOutput `
#69455 opened
Jun 10, 2024
218 Unresolved conversations
Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.
-
TF 2.16.1 Fails to work with GPUs
#63362 commented on
Jun 15, 2024 • 9 new comments -
Compilation of mlir:tf-opt fails with error "The repository '@llvm_zlib' could not be resolved"
#69367 commented on
Jun 14, 2024 • 5 new comments -
NumPy 2.0 support
#67291 commented on
Jun 15, 2024 • 4 new comments -
[RNN] Keras LSTM converted to "While" OPs with hidden states manipulation - TFLite
#62775 commented on
Jun 13, 2024 • 4 new comments -
TypeError: this __dict__ descriptor does not support '_DictWrapper' objects
#62217 commented on
Jun 14, 2024 • 4 new comments -
Tensorflow lite in Android App
#67699 commented on
Jun 13, 2024 • 3 new comments -
Wrong quantized_dimension (axis) when "per-channel" quantization
#66081 commented on
Jun 12, 2024 • 3 new comments -
Instructions to install Tensorflow
#68394 commented on
Jun 13, 2024 • 3 new comments -
Why tf.data.Dataset.choose_from_datasets() chooses only one element from dataset of size-element 5, I want to unite with other dataset of size-element 5 the same. If I want to merge dataset with all their elements and get <ChooseDataset ...> with 10 elements inside
#67327 commented on
Jun 10, 2024 • 3 new comments -
CI build gives a command not found error on /install/install_pip_packages.sh
#62645 commented on
Jun 14, 2024 • 3 new comments -
MobileNetV3 quantization
#69311 commented on
Jun 11, 2024 • 3 new comments -
There is no target called wheel
#68702 commented on
Jun 15, 2024 • 3 new comments -
Doc(Transfer learning and fine-tuning) is quite different from real executive result.
#66696 commented on
Jun 12, 2024 • 2 new comments -
Tensorflow profiler is not showing anything. Gives "No profile data was found" text on selecting Profile in Tensorboard
#61212 commented on
Jun 10, 2024 • 2 new comments -
armeabi-v7a assembler error
#59970 commented on
Jun 11, 2024 • 2 new comments -
tf.truncatediv does not support float/complex tensor
#62071 commented on
Jun 12, 2024 • 2 new comments -
Local rendezvous is aborting with status: OUT_OF_RANGE: End of sequence when call some methods of `tf.data`
#68593 commented on
Jun 15, 2024 • 2 new comments -
The signatures in SavedModel do not contain serving_default when a subclass of the keras model has multiple inputs
#69307 commented on
Jun 11, 2024 • 2 new comments -
Model weights cannot be saved
#68467 commented on
Jun 14, 2024 • 2 new comments -
Memory leak in forward pass (e.g., of ResNet50 model) with TensorFlow 2.12.0 and Python 3.11
#60131 commented on
Jun 11, 2024 • 2 new comments -
Could not load library cudnn_cnn_infer64_8.dll. Error code 126
#63702 commented on
Jun 11, 2024 • 2 new comments -
[RNN] GRU conversion/performance issues on CPU on Windows machines
#57977 commented on
Jun 14, 2024 • 2 new comments -
Tensorflow version 2.16.1 has retracing problem for keras.model.train_on_batch().
#67033 commented on
Jun 14, 2024 • 2 new comments -
TensorFlowLiteSelectTfOps - Compile error
#67748 commented on
Jun 14, 2024 • 2 new comments -
Model containing LSTM does not run after conversion using ACTIVATIONS_INT16_WEIGHTS_INT8 quantization
#60884 commented on
Jun 13, 2024 • 2 new comments -
Too slow while fetching @llvm-raw repos while building tensorflow from Source
#64878 commented on
Jun 13, 2024 • 1 new comment -
`Bias` fails to broadcast in the context of `matmul` in tf lite model
#60929 commented on
Jun 13, 2024 • 1 new comment -
TFLite model produces wrong output after fusion optimization
#61967 commented on
Jun 13, 2024 • 1 new comment -
Cannot find any way to install tensorflow<=2.15.0
#66517 commented on
Jun 13, 2024 • 1 new comment -
Segmentation fault (core dumped) in `tf.raw_ops.FractionalMaxPoolGrad`
#66760 commented on
Jun 13, 2024 • 1 new comment -
output_padding argument in Conv1DTranspose
#68505 commented on
Jun 13, 2024 • 1 new comment -
PNG warning
#62907 commented on
Jun 13, 2024 • 1 new comment -
TypeError: Expected int32, got 1e-07 of type 'float' instead.
#68959 commented on
Jun 13, 2024 • 1 new comment -
Missing legal value check for groups parameter of tf.keras.layers.Conv1D
#69101 commented on
Jun 13, 2024 • 1 new comment -
GPU install error
#60144 commented on
Jun 13, 2024 • 1 new comment -
Working code broke after deploying to new installation. ValueError: When using `stateful=True` in a RNN, the batch size must be static. Found dynamic batch size: sequence.shape=(None, xx, xx)
#64061 commented on
Jun 12, 2024 • 1 new comment -
A checker is needed for inputs of Conv layers.
#65214 commented on
Jun 12, 2024 • 1 new comment -
tf.raw_ops.UnicodeEncode: Segmentation fault (core dumped)
#63379 commented on
Jun 12, 2024 • 1 new comment -
`tf.raw_ops.Conv2DBackpropInput` aborts due to lack of input check
#62950 commented on
Jun 12, 2024 • 1 new comment -
tensorflow.org/versions points to latest version for 2.11 onwards
#62389 commented on
Jun 12, 2024 • 1 new comment -
Different behaviors of raw_ops.Sigmoid can be observed when jitcompiled=true.
#62212 commented on
Jun 12, 2024 • 1 new comment -
Different Behavior of tf.raw_ops.Cosh with jit_compile=True
#62236 commented on
Jun 12, 2024 • 1 new comment -
Internal quantize ops don't match external quantization
#62530 commented on
Jun 12, 2024 • 1 new comment -
C++ API `SparseApplyAdadelta` segfaults due to lack of shape check
#62978 commented on
Jun 12, 2024 • 1 new comment -
PadV2 constant_values tensor not quantized using 16x8 quantization mode
#62499 commented on
Jun 12, 2024 • 1 new comment -
`Check failed` in `tf.transpose`, `tf.raw_ops.Transpose` and `tf.compat.v1.transpose` when the values of `perm` have negative numbers.
#65649 commented on
Jun 12, 2024 • 1 new comment -
C++ API `DenseBincount` violates assertion in shape inference step
#63068 commented on
Jun 12, 2024 • 1 new comment -
Model not learning when using Dataset.from_generator() instead of Dataset.from_tensor_slices()
#53284 commented on
Jun 12, 2024 • 1 new comment -
TensorFlow Lite Inference Crash with `tf.reverse(x, axis=[])`
#62679 commented on
Jun 12, 2024 • 1 new comment -
[RNN] TFLite converter segfaults with GRU models
#62281 commented on
Jun 12, 2024 • 1 new comment -
TFLite model with `l2_normalize(tf.transpose(x))` produces wrong outputs
#61968 commented on
Jun 12, 2024 • 1 new comment -
Module Not Found: yggdrasil_decision_forests.model.gradient_boosted_tree
#69200 commented on
Jun 13, 2024 • 1 new comment -
Support legalization of tf.SplitV op for dynamic shapes
#63026 commented on
Jun 11, 2024 • 1 new comment -
Add support for dataset to pandas dataframe
#43487 commented on
Jun 15, 2024 • 1 new comment -
DLL
#69163 commented on
Jun 15, 2024 • 1 new comment -
TFLiteConverter adds (de)quantization blocks before and after operations on a weight variable
#59390 commented on
Jun 14, 2024 • 1 new comment -
TFLite converter does not support 4-dimensional input for dense operators
#60427 commented on
Jun 14, 2024 • 1 new comment -
Complex dtype input for keras layer in tf2.16+
#65306 commented on
Jun 14, 2024 • 1 new comment -
IntegerLookup layer performance issue when inited with vocabulary
#65610 commented on
Jun 14, 2024 • 1 new comment -
tf.data filter dataset too slow
#67330 commented on
Jun 14, 2024 • 1 new comment -
issue with loss_weights parameter of model.compile() , when model returns multiple output
#67405 commented on
Jun 14, 2024 • 1 new comment -
ValueError: as_list() is not defined on an unknown TensorShape. during training
#68217 commented on
Jun 14, 2024 • 1 new comment -
segmentation fault when tf.histogram_fixed_width receives large `value_range` and `nbins` on CPU mode
#68836 commented on
Jun 14, 2024 • 1 new comment -
__add__ with floating point values
#68923 commented on
Jun 14, 2024 • 1 new comment -
BatchToSpaceND and SpaceToBatchND ERROR_GPU_NOT_COMPATIBLE
#59870 commented on
Jun 14, 2024 • 1 new comment -
Efficientnet B7 classification conversion from tf to tflite fails tflite imagenet evaluation test
#60053 commented on
Jun 14, 2024 • 1 new comment -
Linking an Android library with TFLite GPU using CMake causes undefined symbol errors
#61312 commented on
Jun 14, 2024 • 1 new comment -
TFLite NNAPI Delegate converts INT8 UnidirectionalSequenceLSTM to incorrect NN operation type
#60234 commented on
Jun 14, 2024 • 1 new comment -
Problems with converted 8-bit TFLite models of CycleGAN and running inference (specially allocating tensors)
#59922 commented on
Jun 14, 2024 • 1 new comment -
Quantised fused custom op
#58190 commented on
Jun 14, 2024 • 1 new comment -
converting LSTM layer to tflite with float16 fails
#61370 commented on
Jun 14, 2024 • 1 new comment -
Utilize GPU for tf 2.15
#69042 commented on
Jun 14, 2024 • 1 new comment -
tf.keras.layers.Dense leads to significant differences between CPU and GPU runs of the model implementation code
#67829 commented on
Jun 14, 2024 • 1 new comment -
`softplus` outputs `inf` for large inputs after converting to lite model
#60892 commented on
Jun 13, 2024 • 1 new comment -
`Unpack` and `concat` wrongly transformed into `reshape` in tflite converter
#60925 commented on
Jun 13, 2024 • 1 new comment -
TF Lite produces wrong graph when tensor broadcasting exists
#61150 commented on
Jun 13, 2024 • 1 new comment -
TF-Lite is 4x slower than Tensorflow on MacOS (and 2x slower in Colab)
#60609 commented on
Jun 13, 2024 • 1 new comment -
Converted tflite file is 30x the size of the original SavedModel
#56075 commented on
Jun 13, 2024 • 1 new comment -
Keras docs source code links point to 404
#61429 commented on
Jun 13, 2024 • 1 new comment -
Please update the links on documentation page, pointing to the new location - moved to /src
#66572 commented on
Jun 13, 2024 • 1 new comment -
Uncompliant tflite model when converting "MultiHeadAttention" layer
#61796 commented on
Jun 13, 2024 • 1 new comment -
TF Lite produces wrong graph with a sequence of tensor reshape operators
#61886 commented on
Jun 13, 2024 • 1 new comment -
ELU int8 model quantized with Dequantize/Quantize stubs
#60789 commented on
Jun 13, 2024 • 1 new comment -
TensorFlow 2.16 / Keras 3 have undocumented breaking API changes
#63792 commented on
Jun 13, 2024 • 1 new comment -
QuantizedOpsTest.testAxis fails on cascade lake CPUs
#49944 commented on
Jun 10, 2024 • 1 new comment -
ERROR: @local_config_cuda//:enable_cuda :: Error loading option @local_config_cuda//:enable_cuda: 'NoneType' value has no field or method 'replace'
#65195 commented on
Jun 10, 2024 • 1 new comment -
TFLite for LSTM: Downscale accumulation from 32-bit to 16-bit before applying to activation
#68670 commented on
Jun 10, 2024 • 1 new comment -
RuntimeError when invoking TFLite INT8 model with tile operation
#67789 commented on
Jun 10, 2024 • 1 new comment -
Conversion failure: tfl.batch_matmul "expected 3 but got 2" (regression since 2.14, worked in 2.13)
#65769 commented on
Jun 10, 2024 • 1 new comment -
TFLite Op type not registered (RegexSplitWithOffsets) in Swift
#65475 commented on
Jun 10, 2024 • 1 new comment -
Running an Integrated Image Segmenter in Java
#69021 commented on
Jun 11, 2024 • 1 new comment -
tflite-runtime 2.11 python wheel for windows
#69020 commented on
Jun 11, 2024 • 1 new comment -
ValueError: `validation_split` is only supported for Tensors or NumPy arrays, found following types in the input: [<class 'int'>]
#68882 commented on
Jun 11, 2024 • 1 new comment -
Need Help with a Softmax Warning in TensorFlow 2.16
#67758 commented on
Jun 11, 2024 • 1 new comment -
tf.image.draw_bounding_boxes: Aborted (core dumped)
#63688 commented on
Jun 11, 2024 • 1 new comment -
tf.keras.layers.PReLU outputs NaN on positive input
#63823 commented on
Jun 11, 2024 • 1 new comment -
Aborted in `tf.reduce_mean` occurs when gpu is not available
#69054 commented on
Jun 11, 2024 • 1 new comment -
MultiWorkerMirrorStrategy Metrics Incorrectly Aggregating
#64471 commented on
Jun 11, 2024 • 1 new comment -
Aborted (core dumped) in `tf.transpose`
#69213 commented on
Jun 11, 2024 • 1 new comment -
Aborted (core dumped) in `tf.raw_ops.QuantizeAndDequantizeV3`
#69220 commented on
Jun 11, 2024 • 1 new comment -
Aborted (core dumped) in `tf.compat.v1.image.draw_bounding_boxes`
#69232 commented on
Jun 11, 2024 • 1 new comment -
TFLite GPUv2: ADD(x, 1e-5) results in severely wrong output
#67216 commented on
Jun 9, 2024 • 1 new comment -
TensorFlow Cuda in Docker under WSL2 not wokring
#68710 commented on
Jun 10, 2024 • 1 new comment -
Request for groups parameter support in Conv2DTranspose/Conv1DTranspose Layer
#69201 commented on
Jun 10, 2024 • 1 new comment -
The documentation for Conv1DTranspose does not state that the CPU does not support dilation rates larger than 1
#69103 commented on
Jun 10, 2024 • 1 new comment -
DXGI format does not support cross-API sharing
#69430 commented on
Jun 10, 2024 • 1 new comment -
Support for 16 bit activations in ExpandDims operation.
#68293 commented on
Jun 10, 2024 • 1 new comment -
Make TensorFlow Lite available as Swift Package Manager package
#44609 commented on
Jun 10, 2024 • 1 new comment -
Can't find libdevice directory ${CUDA_DIR}/nvvm/libdevice
#56927 commented on
Jun 10, 2024 • 1 new comment -
Title: TensorFlow/Keras Integration Error: A KerasTensor cannot be used as input to a TensorFlow function
#69340 commented on
Jun 10, 2024 • 1 new comment -
Unable to build TensorFlowLite GPU Delegate for Android
#69252 commented on
Jun 10, 2024 • 1 new comment -
GPU MaxPool gradient ops do not yet have a deterministic XLA implementation
#69417 commented on
Jun 10, 2024 • 1 new comment -
Couldn't resolve TF-TRT Warning: Could not find TensorRT
#68335 commented on
Jun 10, 2024 • 1 new comment -
Failing Tensorflow unit tests for BF16 hardware
#65988 commented on
Jun 10, 2024 • 1 new comment -
Documentation for `tf.linalg.set_diag()` is missing return information
#67255 commented on
Jun 10, 2024 • 1 new comment -
tf.tensor_scatter_nd_update lead to a program abortion when receiving a 3d indices
#63575 commented on
Jun 10, 2024 • 1 new comment -
`Check failed` in `tf.raw_ops.TensorScatterMin` and `tf.tensor_scatter_nd_min` when the rank of `indices` > 2.
#65669 commented on
Jun 10, 2024 • 1 new comment -
tf.tensor_scatter_nd_update: Aborted (core dumped)
#63375 commented on
Jun 10, 2024 • 1 new comment -
RaggedTensors should have a 'name' attribute
#56819 commented on
Jun 10, 2024 • 1 new comment -
tf.linalg.normalize generates wrong output in tflite version running on mobile GPU
#64922 commented on
Jun 11, 2024 • 1 new comment -
tf.keras.utils.plot_model doesn't work
#65331 commented on
Jun 11, 2024 • 1 new comment -
TFLite 2.16.1 conversions fail with "AttributeError: 'Sequential' object has no attribute '_get_save_spec'"
#63867 commented on
Jun 11, 2024 • 1 new comment -
What is the effect of TF_GUARDED_BY(mu) for variables like Tensor?
#64845 commented on
Jun 11, 2024 • 1 new comment -
Failure in convert Gemma 2B models to TfLite
#63025 commented on
Jun 11, 2024 • 1 new comment -
inf outputs with an OpenCL delegate for a pattern with a sequence of Dense/FullyConnected layers
#62908 commented on
Jun 11, 2024 • 1 new comment -
Performance differences from TFLite delegate and Apple CoreML API
#62884 commented on
Jun 11, 2024 • 1 new comment -
Having non-converted operations, even for simplest models
#62855 commented on
Jun 11, 2024 • 1 new comment -
Could not load library cudnn_cnn_infer64_8.dll. Error code 126
#63455 commented on
Jun 11, 2024 • 1 new comment -
tflite RNN model invoke failed with "num_input_elements != num_output_elements (4288 != 64)Node number 18 (RESHAPE) failed to prepare.Node number 5 (WHILE) failed to invoke."
#62840 commented on
Jun 11, 2024 • 1 new comment -
Add support for TensorRT 10
#66473 commented on
Jun 12, 2024 • 1 new comment -
tf.raw_ops.ResourceApplyGradientDescent: Aborted (core dumped)
#63695 commented on
Jun 11, 2024 • 1 new comment -
Apple Silicon, building pip package ... clang: error: linker command failed with exit code 1
#67473 commented on
Jun 12, 2024 • 1 new comment -
tf.audio.decode_wav: Aborted (core dumped)
#63687 commented on
Jun 12, 2024 • 1 new comment -
Unable to Force-load TensorFlowLiteSelectTfOps.framework, created with Selective Build, in iOS
#67790 commented on
Jun 12, 2024 • 1 new comment -
tflite model maker not install
#69431 commented on
Jun 11, 2024 • 1 new comment -
Aborted (core dumped) in `tf.raw_ops.ResourceApplyFtrl/tf.raw_ops.ResourceApplyFtrlV2/tf.raw_ops.ResourceSparseApplyFtrl/tf.raw_ops.ResourceSparseApplyFtrlV2`
#69278 commented on
Jun 11, 2024 • 1 new comment -
tf.random.normal() causes RAM usage to keep growing
#62203 commented on
Jun 11, 2024 • 1 new comment -
Tensorflow 2.11 and 2.14 Memory Issue
#60469 commented on
Jun 11, 2024 • 1 new comment -
Aborted (core dumped) in `tf.raw_ops.ResourceApplyRMSProp/tf.raw_ops.ResourceSparseApplyRMSProp`
#69281 commented on
Jun 11, 2024 • 1 new comment -
Aborted (core dumped) in `tf.raw_ops.ResourceApplyAdagrad/tf.raw_ops.ResourceApplyAdagradDA/tf.raw_ops.ResourceApplyAdagradV2`
#69285 commented on
Jun 11, 2024 • 1 new comment -
PR #13372: [GPU] Fix autotuner_util_test.
#69146 commented on
Jun 10, 2024 • 0 new comments -
[tsl] logging_test: test LOG/VLOG/VLOG_IS_ON and associated flags/envvars
#69416 commented on
Jun 12, 2024 • 0 new comments -
May fix checkfail in Gatherv2 Op.
#63054 commented on
Jun 13, 2024 • 0 new comments -
Fix Checkfail in raw_ops.DecodeAndCropJpeg
#63071 commented on
Jun 13, 2024 • 0 new comments -
Fix checkfail in ThreadUnsafeUnigramCandidateSampler
#63295 commented on
Jun 13, 2024 • 0 new comments -
update cpuinfo
#63850 commented on
Jun 14, 2024 • 0 new comments -
Typos are fixed in quantization_debugger.ipynb
#63959 commented on
Jun 13, 2024 • 0 new comments -
Introduce hermetic CUDA in Google ML projects.
#64130 commented on
Jun 12, 2024 • 0 new comments -
Unregister complex dtypes for Round OP
#65396 commented on
Jun 13, 2024 • 0 new comments -
Remove redundant std::optional: the field is always present.
#69393 commented on
Jun 10, 2024 • 0 new comments -
Update multinomial_op logits invalid arguments check description
#64651 commented on
Jun 13, 2024 • 0 new comments -
[tf.data] Add `synchronous` parameter to `map`.
#64712 commented on
Jun 11, 2024 • 0 new comments -
Upgrade to support and default to clang 18 for the OSS compiler
#65084 commented on
Jun 12, 2024 • 0 new comments -
Clean up TF deps internal:common for legalize_common
#69415 commented on
Jun 10, 2024 • 0 new comments -
Fix xprofilez integration_tests:xprofilez_handler_gpu_test fail
#69166 commented on
Jun 11, 2024 • 0 new comments -
Add more pattern to HloUnstacker.
#69173 commented on
Jun 11, 2024 • 0 new comments -
Add the batch_padding_policy attribute to BatchFunction.
#69049 commented on
Jun 11, 2024 • 0 new comments -
Move `::mlir::lite::QuantizeWeights` from `TfLiteStatus` to `absl::Status`.
#68996 commented on
Jun 14, 2024 • 0 new comments -
Return absl::Status not TFLiteStatus from ::tflite::optimize::QuantizeWeights.
#68994 commented on
Jun 14, 2024 • 0 new comments -
Add additional overloaded versions of the BufferFromHostLiteral function in PjRtClient which take a device_layout parameter.
#68827 commented on
Jun 10, 2024 • 0 new comments -
[tsl] forward TSL logging to Absl logging
#69179 commented on
Jun 10, 2024 • 0 new comments -
Store device manager within TFRTSession instead of graph_executor, so the device manager is created when the TFRTSession is initialized, instead of when TFRTSession is created.
#68753 commented on
Jun 11, 2024 • 0 new comments -
Clean up TF deps tf_to_xla_attribute_utils
#69197 commented on
Jun 13, 2024 • 0 new comments -
PR #13108: [GPU] Shard GEMM fusion autotuning across multiple compilation processes.
#69233 commented on
Jun 10, 2024 • 0 new comments -
Adding some VLOGs to async collective creator and merger for debugging.
#69239 commented on
Jun 11, 2024 • 0 new comments -
Add support for float8_e4m3fn and float8_e5m2 matmuls in HLO evalulator and XLA CPU
#69245 commented on
Jun 14, 2024 • 0 new comments -
[tf] Upgrade Abseil to LTS branch from Jan 2024, Patch 20240116_2
#69255 commented on
Jun 15, 2024 • 0 new comments -
[tflite] add missing include for absl::StrCat
#69264 commented on
Jun 15, 2024 • 0 new comments -
Updating auto generated .pyi files when updating mypy to v1.10.0.
#69318 commented on
Jun 10, 2024 • 0 new comments -
Add the support for other batch policies into SharedBatchScheduler.
#68612 commented on
Jun 10, 2024 • 0 new comments -
Introduce the support for the greedy (kMinimizeTpuCostPerRequest) batch policy.
#68602 commented on
Jun 10, 2024 • 0 new comments -
Layout optimizer and transposing scalars
#68488 commented on
Jun 13, 2024 • 0 new comments -
Introduce utility function GetPrevAllowedBatchSize.
#66624 commented on
Jun 10, 2024 • 0 new comments -
Introduce the MaybeBatchDown helper method, only supporting kBatchDown at this time.
#66623 commented on
Jun 10, 2024 • 0 new comments -
Keeps kv_store in ifrt::PjRtClient if it is set in the CreateOption.
#69327 commented on
Jun 12, 2024 • 0 new comments -
Introduce the --tensorflow_batch_padding_policy flag.
#66620 commented on
Jun 10, 2024 • 0 new comments -
Move `tsl/framework` to `xla/tsl/framework`
#66527 commented on
Jun 14, 2024 • 0 new comments -
[oneDNN] QuantizeV2 with bfloat16 Input
#66085 commented on
Jun 13, 2024 • 0 new comments -
New Features for TFLite Delegates accuracy and correctness tools
#62937 commented on
Jun 13, 2024 • 0 new comments -
Aborted (core dumped) in `tf.raw_ops.ResourceSparseApplyAdagrad/tf.raw_ops.ResourceSparseApplyAdagradDA/tf.raw_ops.ResourceSparseApplyAdagradV2`
#69284 commented on
Jun 11, 2024 • 0 new comments -
Crash in `tf.raw_ops.ResizeNearestNeighbor/ResizeNearestNeighborGrad/ResizeArea/ResizeBicubic/ResizeBilinear`
#69322 commented on
Jun 11, 2024 • 0 new comments -
Build Tensorflow version that detects CPU instruction set at runtime and lights-up/down
#25590 commented on
Jun 11, 2024 • 0 new comments -
How can I exit the XLAControlFlowContext when inside a jit_compile tf.function? Exit() function take no effect.
#63632 commented on
Jun 11, 2024 • 0 new comments -
Aborted (core dumped) in `tf.raw_ops.ResourceApplyKerasMomentum/tf.raw_ops.ResourceSparseApplyKerasMomentum`
#69279 commented on
Jun 11, 2024 • 0 new comments -
Aborted (core dumped) in `tf.raw_ops.ResourceSparseApplyAdadelta/tf.raw_ops.ResourceApplyAdadelta`
#69283 commented on
Jun 11, 2024 • 0 new comments -
Aborted (core dumped) in `tf.raw_ops.ResourceApplyCenteredRMSProp/tf.raw_ops.ResourceSparseApplyCenteredRMSProp`
#69286 commented on
Jun 11, 2024 • 0 new comments -
Not getting the same result when using .tflite in C and Python.
#65935 commented on
Jun 11, 2024 • 0 new comments -
TF Lite. Cmake. Latest git repo fails to compile from source on windows.
#69036 commented on
Jun 11, 2024 • 0 new comments -
GPUv2 numerical inaccuracy in simple Add + Mul
#66740 commented on
Jun 11, 2024 • 0 new comments -
TFLiteConverter produces model that doesn't conform to GPUv2 (TfLiteGpuDelegate Init: FULLY_CONNECTED: Amount of input channels should match weights width)
#66729 commented on
Jun 11, 2024 • 0 new comments -
Op support request: Matmul with constant left hand side
#66727 commented on
Jun 11, 2024 • 0 new comments -
GPUv2 segfaults on split-head attention CLIP model
#66721 commented on
Jun 11, 2024 • 0 new comments -
Segmentation fault when using tflite_model_maker searcher.TextDataLoader.create(EmbeddingModel, l2_normalize=True)
#65409 commented on
Jun 11, 2024 • 0 new comments -
TFLite Interpreter fails to load fp32/ fp16 model on iPhone with CoreML or Metal Delegate in Swift
#62360 commented on
Jun 12, 2024 • 0 new comments -
TFlite model signature lost after populating with metadata
#62620 commented on
Jun 12, 2024 • 0 new comments -
Aborted (core dumped) in `tf.raw_ops.NearestNeighbors`
#66765 commented on
Jun 12, 2024 • 0 new comments -
Aborted (core dumped) with `tf.raw_ops.LoadAndRemapMatrix`
#64655 commented on
Jun 12, 2024 • 0 new comments -
Validate argument minvalue of tf.random.uniform
#62807 commented on
Jun 13, 2024 • 0 new comments -
[XLA][StreamExecutor] add empty implementation for host stream, avoid…
#61888 commented on
Jun 13, 2024 • 0 new comments -
Change TFL_MINIMUM_OS_VERSION to build TensorFlowLiteCMetal_framework on XCode 14.3
#61174 commented on
Jun 13, 2024 • 0 new comments -
Go: add support for empty tags-set when loading saved model
#60056 commented on
Jun 13, 2024 • 0 new comments -
[TFLite] Add support for int8 quantized DivOp
#59937 commented on
Jun 13, 2024 • 0 new comments -
Fix endianness issues in arithmetic_optimizer_test.cc tests
#59851 commented on
Jun 13, 2024 • 0 new comments -
Update - Image Classification
#57046 commented on
Jun 13, 2024 • 0 new comments -
Fix cuDNN LSTM implementation selection with LoadSavedModel C++ API.
#56525 commented on
Jun 13, 2024 • 0 new comments -
TF-TRT Warning: Could not find TensorRT
#64809 commented on
Jun 10, 2024 • 0 new comments -
Aborted (core dumped) in `tf.raw_ops.ResourceApplyAdaMax/tf.raw_ops.ResourceApplyAdam/tf.raw_ops.ResourceApplyAdamWithAmsgrad`
#69289 commented on
Jun 10, 2024 • 0 new comments -
`Check failed` in `tf.raw_ops.TensorScatterAdd` and `tf.tensor_scatter_nd_add` when the rank of `indices` > 2.
#65671 commented on
Jun 14, 2024 • 0 new comments -
`Check Failed` in `tf.raw_ops.FakeQuantWithMinMaxVarsPerChannel` and `tf.quantization.fake_quant_with_min_max_vars_per_channel` when the input of `inputs` is scalar.
#65728 commented on
Jun 14, 2024 • 0 new comments -
GlobalAveragePooling1D fails with empty inputs and a mask
#67023 commented on
Jun 14, 2024 • 0 new comments -
Numerical precision issue of operators selu, leakyRelu, softplus and their corresponding backward operators on Bfloat16 vs float32
#67440 commented on
Jun 14, 2024 • 0 new comments -
Strange finding: When the global seed and @tf.function decorator are used, the random sampling values of the two adjacent periods are equal
#68215 commented on
Jun 14, 2024 • 0 new comments -
errors in the descriptions of the parameters in the documentation for tf.keras.layers.Conv2DTranspose
#69098 commented on
Jun 10, 2024 • 0 new comments -
No such file or directory: 'patchelf' while compiling from source
#68247 commented on
Jun 14, 2024 • 0 new comments -
TensorRT no longer has NvUtils.h - build from source is failing
#68360 commented on
Jun 14, 2024 • 0 new comments -
TFLite ConvTranspose3D implemented typo
#68319 commented on
Jun 14, 2024 • 0 new comments