Insights: tensorflow/tensorflow

20 Pull requests merged by 15 people

• tanh float32 over 1.

  #55394 merged Dec 22, 2022

• Updated Documentation on CoreML delegate

  #57578 merged Dec 22, 2022

• Update api_def_Angle.pbtxt

  #58685 merged Dec 21, 2022

• [TOSA]Add wrapping of TFL CustomOp attributes to TOSA

  #58827 merged Dec 26, 2022

• [NVIDIA TF] Enable BF16 ArgMax/ArgMin

  #58836 merged Dec 24, 2022

• [NVIDIA TF] Enable BF16 for Dilation

  #58838 merged Dec 23, 2022

• [NVIDIA TF] Enable BF16 TensorArray scatter, split, unpack

  #58841 merged Dec 22, 2022

• Select a convolution algorithm that does not use tensor cores when tensor cores are disabled and autotuning is disabled.

  #58847 merged Dec 20, 2022

• [TF-TRT] Fix AddN for const layer inputs in non implicit batch cases.

  #58895 merged Dec 20, 2022

• Fix buffer overflow in summary_db_writer

  #58913 merged Dec 27, 2022

• Fix copy-paste error with possible null dereference

  #58916 merged Dec 26, 2022

• Fix possible null dereference in xla_launch_util.cc

  #58922 merged Dec 27, 2022

• Re-fix nvidia; remove "devel" images

  #58948 merged Dec 20, 2022

• LHLO Translation Tests for GEMM Custom Calls

  #58949 merged Dec 23, 2022

• [xla:gpu:runtime] Fix integer overflow in GetDeviceBufferPairs

  #58959 merged Dec 23, 2022

• Remove tf.contrib mentions.

  #58963 merged Dec 26, 2022

• Update the RBE images to the latest container versions

  #58968 merged Dec 21, 2022

• [NVIDIA TF] Fix missing cuda headers in pip wheel

  #58976 merged Dec 26, 2022

• [XLA:GPU] Add basic NVTX support to XLA Runtime

  #58977 merged Dec 26, 2022

• [Linaro:ARM_CI] Drop Python 3.7 build as not supported

  #58984 merged Dec 23, 2022

6 Pull requests opened by 6 people

• Fix memory leak in xla::SlowOperationAlarm

  #58970 opened Dec 21, 2022

• Removes unnecessary libgomp linkopt from oneDNN build on AArch64

  #58974 opened Dec 21, 2022

• [tflite] Add rank checking for MEAN, ADD, MUL, and DIV

  #58978 opened Dec 22, 2022

• Checks for integer overflow in function tf.convert_to_tensor() ---->

  #58996 opened Dec 24, 2022

• Fix a check fail in GRUBlockCell

  #58998 opened Dec 25, 2022

• build: fix CMake rules for the MHLOTOSATransforms target

  #59019 opened Dec 26, 2022

72 Issues closed by 18 people

• RuntimeError: Quantization not yet supported for op: 'DEQUANTIZE', after QAT quantize_apply()

  #45328 closed Dec 27, 2022

• Problems to convert a tensorflow model to TFLite

  #44918 closed Dec 27, 2022

• NNAPI does not accept the reshape operation with the -1 parameter

  #25661 closed Dec 27, 2022

• Support for keepdims and padding in tf.boolean_mask or tf.ragged.boolean_mask

  #59024 closed Dec 27, 2022

• tf.truncatediv not support float, but doc support.

  #59018 closed Dec 27, 2022

• error: 'tf.Conv2D' op is neither a custom op nor a flex op

  #59017 closed Dec 27, 2022

• Select TensorFlow op（FlexTensorArrayV3）is not supported by this interpreter

  #58833 closed Dec 27, 2022

• TensorFlow Lite Model Maker has version conflict during install

  #58680 closed Dec 27, 2022

• How to reduce power consumption for gpu delegate on android

  #58855 closed Dec 26, 2022

• Crash when running tf.test.create_local_cluster

  #59011 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.random_grad.add_leading_unit_dimensions

  #59015 closed Dec 26, 2022

• Segmentation fault when running tensorflow.python.ops.nn_impl.fused_batch_norm

  #59014 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.gen_array_ops.pad_v2

  #59005 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.gen_data_flow_ops.record_input

  #59012 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.gen_bitwise_ops.population_count

  #59010 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.gen_array_ops.upper_bound

  #59009 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.gen_collective_ops.collective_gather

  #59008 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.ragged.ragged_string_ops.ngrams

  #59007 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.linalg.linalg_impl.matrix_exponential

  #59006 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.collective_ops.all_gather

  #59004 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.gen_image_ops.extract_glimpse_v2

  #59003 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.gen_linalg_ops.cholesky

  #59002 closed Dec 26, 2022

• Segmentation fault when running tensorflow.python.ops.gen_nn_ops.fused_batch_norm_v3

  #59001 closed Dec 26, 2022

• Crash when running tensorflow.python.ops.gen_array_ops.quantize_v2

  #59000 closed Dec 26, 2022

• Missing zip file

  #58991 closed Dec 26, 2022

• __init__() got an unexpected keyword argument 'reduction'

  #58972 closed Dec 26, 2022

• tf.Module does not update state when a trackable attribute is reassigned a nontrackable

  #37806 closed Dec 25, 2022

• Native support for StridedSlice in 6D and Transpose in 7D

  #53702 closed Dec 24, 2022

• Pull Requests: Map needed for request workflow

  #33802 closed Dec 24, 2022

• compile tensorflow-lite-select-tf-ops no error show,but exit

  #58873 closed Dec 24, 2022

• gpu metal lack ”inference_context_generated.h“

  #58609 closed Dec 23, 2022

• Segment Fault in AudioSpectrogram

  #58105 closed Dec 23, 2022

• Crash due to illegal CUDA memory access in BiasAddGrad

  #58162 closed Dec 23, 2022

• Deploy yolo5 model into TensorFlow Lite Object Detection Android

  #58812 closed Dec 23, 2022

• ValueError: The same saveable will be restored with two names: layer_with_weights-1/_table/.ATTRIBUTES/table

  #44005 closed Dec 23, 2022

• Quantized MobileBERT and ALBERT models performing very poorly

  #44306 closed Dec 23, 2022

• Tensorflow gpu does not work with RTX 3000 series card.

  #45285 closed Dec 23, 2022

• TypeError: Could not build a TypeSpec for KerasTensor

  #58983 closed Dec 23, 2022

• Tensroflow recognizes only one GPU

  #58935 closed Dec 22, 2022

• TFRecord guide doesn't show how to serialize and parse tensors

  #29877 closed Dec 22, 2022

• 'utf-8' codec can't decode byte 0xcd in position 133: invalid continuation byte

  #58798 closed Dec 22, 2022

• un-deprecate `write_grads` for `fit`

  #31173 closed Dec 22, 2022

• Training takes much more time after load saved model(pretrained model)

  #58490 closed Dec 22, 2022

• unable to save function, WeightNormalization with Conv1DTranspose error

  #58945 closed Dec 22, 2022

• ModuleNotFoundError: No module named 'tensorflow.compat'

  #58808 closed Dec 22, 2022

• Did not find a parser for CONV_2D with latest repo download

  #42972 closed Dec 22, 2022

• running TF operations on mobile GPU - error conversion to dynamic graph yields issues while running on mobile GPU

  #43202 closed Dec 22, 2022

• distributed_dataset tensorflow.python.framework.errors_impl.UnavailableError: Socket closed

  #43362 closed Dec 22, 2022

• [ROCm] Incorrect path to clang.

  #43468 closed Dec 22, 2022

• Debugger V2 not working. Invalid argument: DebugNumericSummaryV2Op requires tensor_id to be less than or equal to (2^53). Given tensor_id:26

  #43608 closed Dec 22, 2022

• tf.keras.Model with multiple outputs, using fit() with a generator dataset passes y to loss with wrong shape

  #43379 closed Dec 22, 2022

• BatchNormalization is not found under `from tensorflow.python.keras.layers`

  #58947 closed Dec 22, 2022

• Unable to build Tensorflow 2.11.0 from source code

  #58953 closed Dec 22, 2022

• A check fail can be triggered in ParameterizedTruncatedNormal

  #58274 closed Dec 21, 2022

• Error in inference when the model is exported with

  #58751 closed Dec 21, 2022

• tf.linalg.matrix_rank has different results with or without @tf.function

  #57959 closed Dec 21, 2022

• GPUs only visible for administrators

  #58957 closed Dec 21, 2022

• release variable memory api ?

  #58487 closed Dec 21, 2022

• when running tflite mode using GPU , the result on AMD is wrong

  #58857 closed Dec 20, 2022

• Computing jacobian for `tf.image.resize` cause Check Failed on cuda

  #57880 closed Dec 20, 2022

• Process crash caused by check failed in `UnsortedSegmentProd`

  #57798 closed Dec 20, 2022

• [Keras] `Conv2DTranspose` constfolding issue with variable batch size.

  #57853 closed Dec 20, 2022

• [TF] `tf.image.resize_with_crop_or_pad` constfolding issue with variable batch size.

  #57898 closed Dec 20, 2022

• Inconsistencies between 2d and 3d ops

  #57835 closed Dec 20, 2022

• tf.config.experimental.enable_tensor_float_32_execution(False) fails with TF_CUDNN_USE_AUTOTUNE=0

  #58846 closed Dec 20, 2022

• Possible bug / Documentation Suggestion - Dataset API + Estimator API does not throw an exception when the input filepath to the dataset is incorrect

  #21335 closed Dec 20, 2022

• ERROR: No matching distribution found for tensorflow==2.13.0

  #58843 closed Dec 20, 2022

• No clear_devices in BestExporter

  #23900 closed Dec 20, 2022

• Optimizer within Estimator computes incorrectly small gradient when used with MirroredStrategy

  #35333 closed Dec 20, 2022

• Debugging predictions step with tf.debugging.experimental.enable_dump_debug_info throws an error.

  #41392 closed Dec 20, 2022

• Windows Build error with --cofig=mkl => ImportError: DLL load failed while importing _pywrap_tensorflow_internal

  #55817 closed Dec 20, 2022

• Unexpected rank of mask computed by tf.keras.layers.Masking

  #48026 closed Dec 20, 2022

31 Issues opened by 31 people

• TFlite 2.11.0 C API no longer compiles due to missing dependency

  #59026 opened Dec 27, 2022

• ERROR: Could not find a version that satisfies the requirement tensorflow==1.15.2

  #59025 opened Dec 27, 2022

• struct.error: unpack_from requires a buffer of at least 4 bytes for unpacking 4 bytes at offset 0 (actual buffer size is 0)

  #59023 opened Dec 27, 2022

• [NVIDIA XLA] large GPU memory overhead observed due to seems un-necessary copy operators introduced by XLA

  #59022 opened Dec 27, 2022

• tensor_util.py

  #59021 opened Dec 26, 2022

• RuntimeError: MetaGraphDef when converting to tflite

  #59020 opened Dec 26, 2022

• Window only link

  #59016 opened Dec 26, 2022

• Issue with CUDA library directory not being detected properly by Tensorflow Ubuntu 22.04

  #59013 opened Dec 25, 2022

• Segmentation fault in tensorflow.python.ops.gen_nn_ops.fused_batch_norm_grad_v3

  #58999 opened Dec 25, 2022

• Inconsistent predict results

  #58997 opened Dec 25, 2022

• `PluggableGraphOptimizer failed: NOT_FOUND: Op type not registered '_CopyFromGpuToHost'`

  #58995 opened Dec 24, 2022

• Missing interger overflow check in TensorFlow

  #58994 opened Dec 24, 2022

• NotImplementedError: Cannot convert a symbolic tf.Tensor (strided_slice:0) to a numpy array. This error may indicate that you're trying to pass a Tensor to a NumPy call, which is not supported.

  #58993 opened Dec 24, 2022

• How Do I Convert input arg tf.string to string in tf.function

  #58992 opened Dec 24, 2022

• NNAPI initialization from cache has hardly any time benefit

  #58990 opened Dec 23, 2022

• By using tf.control_dependencies in my code, the custom_ops will not be execute

  #58989 opened Dec 23, 2022

• NOT_FOUND: could not find registered platform with id

  #58988 opened Dec 22, 2022

• Unsuported fullyconnected operator type by tflite

  #58987 opened Dec 22, 2022

• Loaded runtime CuDNN library: 8.0.5 but source was compiled with: 8.1.0

  #58986 opened Dec 22, 2022

• Windows C API package is missing files

  #58985 opened Dec 22, 2022

• pip subprocess to install build dependencies did not run successfully

  #58982 opened Dec 22, 2022

• with `with strategy.scope():` BERT output loses it's shape

  #58981 opened Dec 22, 2022

• SavedModel load for tags { serve }; Status: fail: Not found: Op type not registered 'TFRA>CuckooHashTableOfTensors' in binary running on machine....

  #58980 opened Dec 22, 2022

• Installing sktime all_extras got compatible issues with tensorflow

  #58975 opened Dec 21, 2022

• cannot deepcopy Optimizer class anymore in tf>=2.11

  #58973 opened Dec 21, 2022

• XLA Collapse Documentation: Example Incorrect

  #58971 opened Dec 21, 2022

• Almost wiped my PC trying to make Tensorboard work on windows Jupyter notebook

  #58969 opened Dec 21, 2022

• Validate on entire validation_data if Dataset has length less than validation_steps

  #58966 opened Dec 20, 2022

• Support for indexing a container of `Layer`s using index from `Dataset` element inside a `tf.function` or `Layer.call`

  #58965 opened Dec 20, 2022

• Unable to use any augmentation layer within `tf.data` with XLA enabled, in `MultiWorkerMirroredStrategy`

  #58961 opened Dec 20, 2022

• Not able to

  #58960 opened Dec 20, 2022

141 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 2.11.0 fails to build in MacOS 13 - XCode 14.1 - issue with ld linker

  #58368 commented on Dec 25, 2022 • 7 new comments

• Modify the data type of the operator parameter under tf.math.

  #58765 commented on Dec 26, 2022 • 5 new comments

• ROCm support for determinism

  #58813 commented on Dec 22, 2022 • 5 new comments

• TensorFlow device is mapped to multiple devices when using tf.estimator models and setting visible device using TF v1 api with TF2.11

  #58952 commented on Dec 22, 2022 • 4 new comments

• AutoCast variable inside tf.function does not cast correctly when saving a model

  #58937 commented on Dec 23, 2022 • 4 new comments

• [NVIDIA XLA] XLA generates element-wise fusion kernel with no data locality for GPU

  #58950 commented on Dec 26, 2022 • 4 new comments

• tf2.11 tf.profiler.experimental doesn't generate trace.json.gz and other files compared with tf2.10

  #58738 commented on Dec 26, 2022 • 4 new comments

• A check fail can be triggered in LSTMBlockCell

  #58270 commented on Dec 21, 2022 • 3 new comments

• /home/work/tensorflow/test/tensorflow-2.4.0/tensorflow/compiler/xla/service/cpu/BUILD:695:1: C++ compilation of rule '//tensorflow/compiler/xla/service/cpu:runtime_matmul' failed (Exit 4)

  #58956 commented on Dec 22, 2022 • 3 new comments

• Dependencies installation failure in rasa due to Tensor flow

  #58939 commented on Dec 22, 2022 • 3 new comments

• XLA tools build broken (//tensorflow/compiler/xla/tools:replay_computation_gpu ) on tensorflow master branch

  #58851 commented on Dec 22, 2022 • 3 new comments

• Tensorflow Lite conversion output vs. Tensorflow output

  #58206 commented on Dec 23, 2022 • 3 new comments

• The input type of the argmin and argmax operator parameters does not match the exception information.

  #58739 commented on Dec 26, 2022 • 3 new comments

• Tensorfllow issue: local installation not working.

  #58859 commented on Dec 27, 2022 • 3 new comments

• attr return values handled differently than dict by tf.distribute.Strategy.run

  #44515 commented on Dec 27, 2022 • 3 new comments

• tf.io.gfile.GFile truncates gzipped files loaded from Google Cloud buckets

  #40801 commented on Dec 27, 2022 • 3 new comments

• [AMD-ZENDNN] Environment variable to enable AMD-ZENDNN plugin

  #58942 commented on Dec 26, 2022 • 3 new comments

• tensorflow.experimental.numpy.lcm giving inconsistent results

  #58955 commented on Dec 21, 2022 • 2 new comments

• hang in recursive call of google::protobuf::DescriptorPool::FindFileByName

  #42035 commented on Dec 21, 2022 • 2 new comments

• Bad download link for Windows GPU only binaries zip file

  #58933 commented on Dec 21, 2022 • 2 new comments

• Missing docker images for 2.5.2 and 2.5.3

  #58958 commented on Dec 22, 2022 • 2 new comments

• set_intersection doesn't work as expectation - tensorflow 1.6.0

  #19832 commented on Dec 22, 2022 • 2 new comments

• //tensorflow/compiler/mlir/lite/tests:const-fold.mlir.test fails on manylinux2014-aarch64

  #56029 commented on Dec 23, 2022 • 2 new comments

• Consider automatic casting rules for promoting dtypes

  #21491 commented on Dec 23, 2022 • 2 new comments

• ComputeH2D is more priority than FeaturesH2D

  #58616 commented on Dec 24, 2022 • 2 new comments

• tf.distribute.MultiWorkerMirroredStrategy with custom model.train_step

  #58938 commented on Dec 26, 2022 • 2 new comments

• import tensorflow is very slow

  #37729 commented on Dec 26, 2022 • 2 new comments

• #undef some macros with generic names to prevent -Wmacro-redefined from being triggered

  #53750 commented on Dec 27, 2022 • 2 new comments

• [NVIDIA TF] Enable BF16 Scan ops

  #58795 commented on Dec 22, 2022 • 2 new comments

• Fix possible null dereference in hlo_legalize_to_hlo

  #58921 commented on Dec 24, 2022 • 2 new comments

• If TMP is not set, then the build system defaults to the wrong path

  #40614 commented on Dec 20, 2022 • 1 new comment

• TensorFlow doesn't use all available memory on NVIDIA GPU

  #56661 commented on Dec 20, 2022 • 1 new comment

• Tensorflow Layer Reshape not working for same size arrays

  #58786 commented on Dec 20, 2022 • 1 new comment

• Test failures with "OpError not raised"

  #45661 commented on Dec 20, 2022 • 1 new comment

• add_weight with None name generate a graph that it's not possible to save by checkpoint manager

  #35241 commented on Dec 21, 2022 • 1 new comment

• OOM out of memory Mask RCNN

  #58849 commented on Dec 21, 2022 • 1 new comment

• How to get single UnidirectionalSequenceRnnOp in tflite model

  #58854 commented on Dec 21, 2022 • 1 new comment

• TF2.9 perf is slower

  #58791 commented on Dec 21, 2022 • 1 new comment

• Suggested Build Params For Older Gen Intel CPU

  #58944 commented on Dec 21, 2022 • 1 new comment

• Build TF 2.11.0 with use ROCM 5.4.0 with GCC 12.2.0 failed, but succesfull build with use ROCM 5.4.0 with GCC 11.3.0

  #58881 commented on Dec 21, 2022 • 1 new comment

• Enabling tf.function(jit_compile=True) causes inconsistent results

  #58752 commented on Dec 21, 2022 • 1 new comment

• The TfLite "split" kernel fails with kTfLiteInt64 input tensor, even though "split_v" and "concatenate" work

  #50484 commented on Dec 21, 2022 • 1 new comment

• tf.Graph.get_tensor_by_name does not work as expected in TF2

  #37428 commented on Dec 21, 2022 • 1 new comment

• Error following installation, help requested

  #58908 commented on Dec 21, 2022 • 1 new comment

• dataset windows don't work in batches

  #44046 commented on Dec 21, 2022 • 1 new comment

• tensorflow installation not working on apple silicon macs

  #58869 commented on Dec 21, 2022 • 1 new comment

• Potential overflow in libhdfs wrapper

  #20086 commented on Dec 21, 2022 • 1 new comment

• Unacceptable framework overhead for huge networks

  #23545 commented on Dec 22, 2022 • 1 new comment

• Remove output tensors 1 - 4 from FusedBatchNorm

  #23143 commented on Dec 22, 2022 • 1 new comment

• C++ low level API documentation

  #22655 commented on Dec 22, 2022 • 1 new comment

• Tensorflow 2.0 tf.name_scope has no effect on weights created by keras functional api

  #27298 commented on Dec 22, 2022 • 1 new comment

• edgetpu_compiler: "Model not Quantized" despite being quantized

  #46806 commented on Dec 22, 2022 • 1 new comment

• Sparsification of huge matrix

  #23754 commented on Dec 22, 2022 • 1 new comment

• TFLite GPU Delegate Fails

  #47168 commented on Dec 22, 2022 • 1 new comment

• WARNING:absl:Found untraced functions such as lstm_cell_2_layer_call_fn, lstm_cell_2_layer_call_and_return_conditional_losses, lstm_cell_2_layer_call_fn, lstm_cell_2_layer_call_and_return_conditional_losses, lstm_cell_2_layer_call_and_return_conditional_losses while saving (showing 5 of 5). These functions will not be directly callable after loading.

  #47554 commented on Dec 22, 2022 • 1 new comment

• Increased memory usage while loading model graph and observed a drop in performance throughput , When upgrade from TF version 2.8.0 to either 2.8.1 or 2.9.1

  #58140 commented on Dec 22, 2022 • 1 new comment

• C API does not appear to register GPU XLA platform correctly

  #50458 commented on Dec 22, 2022 • 1 new comment

• Convert both model and post processing to tflite model

  #58874 commented on Dec 22, 2022 • 1 new comment

• TF Record, Example, and SeqeunceExample inconsistency(?) / Documentation request

  #22092 commented on Dec 22, 2022 • 1 new comment

• CudaRoot() during compilation should not be used

  #23783 commented on Dec 22, 2022 • 1 new comment

• TF 2.11 on redhat Could not load dynamic library and This TensorFlow binary is optimized.....

  #58824 commented on Dec 23, 2022 • 1 new comment

• tf.custom_gradient with multiple input and output

  #58941 commented on Dec 23, 2022 • 1 new comment

• Error during Inference of LSTM Tflite model

  #55065 commented on Dec 23, 2022 • 1 new comment

• C API installation instructions don't work for MacOS

  #58901 commented on Dec 23, 2022 • 1 new comment

• Tensorflow registering GPU with a lot less memory than available

  #55818 commented on Dec 24, 2022 • 1 new comment

• InaccessibleTensorError in graph mode with for loop itteration (different sized tensors) ( tf.image.resize)

  #37444 commented on Dec 24, 2022 • 1 new comment

• TensorRT converted models are 2-3 times slower in TF 2.10.0 in comparison to TF 2.9.2

  #58248 commented on Dec 24, 2022 • 1 new comment

• Keras model training is slow without "tf.compat.v1.disable_eager_execution()"

  #57645 commented on Dec 25, 2022 • 1 new comment

• CollectiveOpV3Kernel did not support group assignments

  #58906 commented on Dec 26, 2022 • 1 new comment

• Using TF-TRT doubles the size of frozen protobuf file

  #22242 commented on Dec 26, 2022 • 1 new comment

• Bulid TF2.4 in arm architecture, report an error（crosstool_wrapper_driver_is_not_gcc failed, NvInferLegacyDims.h: No such file or directory）

  #58883 commented on Dec 26, 2022 • 1 new comment

• [TF 2.0] tf.assert_equal([], [1.0]) doesn't raise error

  #32082 commented on Dec 26, 2022 • 1 new comment

• Invalid argument in example

  #58621 commented on Dec 26, 2022 • 1 new comment

• Reproducible dataset order after shuffle from checkpoint

  #58925 commented on Dec 26, 2022 • 1 new comment

• Wrong layer name in exception message

  #58926 commented on Dec 27, 2022 • 1 new comment

• CMake build system Tensorflow Lite v2.11.0 is broken

  #58636 commented on Dec 27, 2022 • 1 new comment

• Graph transform tool Node Quantization gives error

  #28276 commented on Dec 27, 2022 • 1 new comment

• tf.random.uniform unexpected behaviour for unknown shape

  #40125 commented on Dec 27, 2022 • 1 new comment

• TPUStrategy.run fails on non-primary thread with "No OpKernel was registered", "TPUReplicatedInput"

  #44487 commented on Dec 27, 2022 • 1 new comment

• Fix XNNPACK build failure with when -mcpu compiler switch is set

  #53449 commented on Dec 21, 2022 • 1 new comment

• Fix typo in GL converter error log

  #56155 commented on Dec 21, 2022 • 1 new comment

• Migration to clang-15

  #57469 commented on Dec 21, 2022 • 1 new comment

• [TFTRT] Check supported precisions before build

  #57591 commented on Dec 26, 2022 • 1 new comment

• Fix ReLUN1To1 fused activation for OpenCL and OpenGL

  #58255 commented on Dec 26, 2022 • 1 new comment

• the priority of H2D

  #58617 commented on Dec 24, 2022 • 1 new comment

• Make fuzztest fuzzer ready for OSS-Fuzz

  #58646 commented on Dec 22, 2022 • 1 new comment

• Updated the deprecated API to new API

  #58672 commented on Dec 27, 2022 • 1 new comment

• Updated image_ops_impl.py

  #58725 commented on Dec 27, 2022 • 1 new comment

• Replace many calls to eval() with self.evaluate()

  #58727 commented on Dec 23, 2022 • 1 new comment

• Updated C++ version and CMake cloning target in microcontroller guide

  #58753 commented on Dec 27, 2022 • 1 new comment

• Update soft NMS factor as per research paper

  #58754 commented on Dec 27, 2022 • 1 new comment

• [NVIDIA TF] Enable BF16 BatchToSpace, DepthToSpace ops

  #58788 commented on Dec 23, 2022 • 1 new comment

• [tosa] Add TFLite reduce all operation to TOSA

  #58800 commented on Dec 26, 2022 • 1 new comment

• [mlir] Revise GetMlirOpNameFromOpCode for custom/while

  #58817 commented on Dec 27, 2022 • 1 new comment

• [NVIDIA XLA] add a new mode that hlo_parser can parse computations without module header.

  #58870 commented on Dec 22, 2022 • 1 new comment

• [ROCm] fixed gpu_kernel_tiling_test on RowReductionCorrectShmemUsage's fusion part

  #58893 commented on Dec 21, 2022 • 1 new comment

• Updating Flatbuffer version to 2.0.8 in flatbuffers.cmake

  #58902 commented on Dec 27, 2022 • 1 new comment

• Added Autoparallel optimizer in config.py

  #58909 commented on Dec 20, 2022 • 1 new comment

• Fix bugs found by static analysis

  #58919 commented on Dec 24, 2022 • 1 new comment

• Upgrading from TF 2.1 to 2.2 gives 12% slowdown and 23% memory increase

  #39434 commented on Dec 20, 2022 • 0 new comments

• Clog Copy-Paste Error

  #58943 commented on Dec 20, 2022 • 0 new comments

• Build on PPC fails due to import of `__subpackages__`

  #58768 commented on Dec 20, 2022 • 0 new comments

• A crash due to check-fail can be triggered in FixedUnigramCandidateSampler

  #58181 commented on Dec 20, 2022 • 0 new comments

• [TF 2.2] Elapsed time of ConcreteFunction becomes shorter when printing loss

  #40659 commented on Dec 21, 2022 • 0 new comments

• `tf.math.negative` negates only the last element in an all-zero 3x3 matrix on CPU

  #58515 commented on Dec 21, 2022 • 0 new comments

• " ValueError: assignment destination is read-only" when using Pandas DataFrame

  #47958 commented on Dec 21, 2022 • 0 new comments

• Infinite loop when using reduce() inside tf.data.Dataset.map()

  #47981 commented on Dec 21, 2022 • 0 new comments

• attr.s decorated keras Layers broken in keras.Sequential

  #47998 commented on Dec 21, 2022 • 0 new comments

• Convert FloorModOp to tosa

  #57520 commented on Dec 21, 2022 • 0 new comments

• Out of memory in some tests due to GPU memory limit confusion

  #45664 commented on Dec 22, 2022 • 0 new comments

• Could not load dynamic library 'libcudart.so.11.0'

  #45930 commented on Dec 22, 2022 • 0 new comments

• Exporting Weights from Tensorflow Keras to C produces a different set of outputs, reduced accuracy in the model.

  #58861 commented on Dec 22, 2022 • 0 new comments

• How to calculate the model's flops when I use tensorflow?

  #58848 commented on Dec 23, 2022 • 0 new comments

• error executing command crosstool_wrapper_driver_is_not_gcc

  #53680 commented on Dec 23, 2022 • 0 new comments

• Jetson Nano build fails with missing "cudnn_version" string (default=7)

  #40784 commented on Dec 23, 2022 • 0 new comments

• Unable to log scalar summaries in XLA

  #39798 commented on Dec 27, 2022 • 0 new comments

• TensorArray issues in XLA

  #42878 commented on Dec 27, 2022 • 0 new comments

• control flow support issue in TensorFlow XLA

  #42876 commented on Dec 27, 2022 • 0 new comments

• Tensor had Inf values (maybe from TensorArray)

  #40228 commented on Dec 27, 2022 • 0 new comments

• learned_unigram_candidate_sampler may enter infinite loop when range_max is a very large number

  #40263 commented on Dec 27, 2022 • 0 new comments

• ```tf.stack``` returns different shapes when run in eager and graph mode

  #42424 commented on Dec 27, 2022 • 0 new comments

• `third_party/py/BUILD.tpl`: add `stub_shebang` for `py_runtime`

  #51599 commented on Dec 22, 2022 • 0 new comments

• fix moments float16 cast bug

  #52217 commented on Dec 26, 2022 • 0 new comments

• [iOS] Add initial xcframework bazel support

  #56789 commented on Dec 20, 2022 • 0 new comments

• [TFTRT] Refactor C++ API

  #57673 commented on Dec 21, 2022 • 0 new comments

• Enable i64 bit indexing for unary MLIR kernels

  #58044 commented on Dec 22, 2022 • 0 new comments

• Represents the in64 bit indexing test for large tensors for Atanh

  #58045 commented on Dec 22, 2022 • 0 new comments

• Sets `--experimental_link_static_libraries_once=true`

  #58209 commented on Dec 23, 2022 • 0 new comments

• bfloat16 GEMM + Bias + ReLU Epilogue Fusion in XLA

  #58360 commented on Dec 20, 2022 • 0 new comments

• [ROCm] Add related hlo tests based on amdgcn for hlo_to_llvm_ir.cc

  #58441 commented on Dec 24, 2022 • 0 new comments

• [TFLite] Remove deprecated gen_lut function

  #58519 commented on Dec 21, 2022 • 0 new comments

• [TFLite] Fix of issue 58050. Give user the option to silence the logging of TfLite completely

  #58590 commented on Dec 22, 2022 • 0 new comments

• Rewrite error messages

  #58665 commented on Dec 20, 2022 • 0 new comments

• Enable i64 bit indexing for binary MLIR kernels

  #58717 commented on Dec 22, 2022 • 0 new comments

• Remove completed TODO bugs.

  #58735 commented on Dec 21, 2022 • 0 new comments

• Fix the endianness issue in lite/kernels:bucketize_test on s390x

  #58803 commented on Dec 21, 2022 • 0 new comments

• Migrate fuzzers to GFT

  #58853 commented on Dec 21, 2022 • 0 new comments

• bincount fuzzer

  #58868 commented on Dec 21, 2022 • 0 new comments

• [oneDNN] Bug-fix: Threadpool interface to LayerNorm.

  #58878 commented on Dec 26, 2022 • 0 new comments

• [oneDNN] Enabled {Conv2D, DepthwiseConv2D}+BiasAdd+FusedHardSwish for FP32/BF16

  #58903 commented on Dec 27, 2022 • 0 new comments

• Fix null dereference from sharding_util

  #58912 commented on Dec 22, 2022 • 0 new comments