tensorflow / tensorflow Public

40 Pull requests merged by 23 people

• Add link to padding notes to depthwise_conv2d doc

  #55175 merged Apr 1, 2022

• Update version numbers for TensorFlow 2.9.0-rc0

  #55458 merged Apr 1, 2022

• Update release notes for TensorFlow 2.9.0

  #55457 merged Apr 1, 2022

• Add error handling to TF_GetInputTensorFromVariable

  #55380 merged Apr 1, 2022

• Fix out size eqn in atrous_conv2d doc

  #55176 merged Apr 1, 2022

• Add ndmin check

  #55430 merged Apr 1, 2022

• Pylint: Revert W + TODO

  #55453 merged Apr 1, 2022

• Create cherry-pick helper GA workflow

  #55459 merged Mar 31, 2022

• [TF-TRT] Slice OP BugFix

  #55353 merged Mar 31, 2022

• Add release notes for manylinux2014 and libcxx ABI change

  #55456 merged Mar 31, 2022

• [TF-TRT] Remove dependency on einsum kernel

  #55320 merged Mar 30, 2022

• Fix pfor defunc issue

  #55350 merged Mar 30, 2022

• Updates mkl_aarch64 build to use oneDNN v2.6.

  #55434 merged Mar 30, 2022

• Fix build windows

  #54311 merged Mar 30, 2022

• Enable pylint warnings

  #55396 merged Mar 30, 2022

• [PluggableDevice] Add TF_IsRefInput

  #55395 merged Mar 30, 2022

• [oneDNN] Fix gru_v2_test_gpu and layer_correctness_test_gpu tests

  #55286 merged Mar 30, 2022

• [PluggableDevice] Add int32 DEVICE_DEFAULT registration for Where

  #54746 merged Mar 30, 2022

• [TF-TRT] Make TF-TRT respect TF32 ON/OFF TF Flag

  #55339 merged Mar 29, 2022

• Enabling Bazelisk to be used when Bazel is not present during install build from source

  #55272 merged Mar 29, 2022

• fix typos

  #55342 merged Mar 29, 2022

• fix broken microcontroller link.

  #55103 merged Mar 29, 2022

• Use `_shape` annotation in ReadVariableOp shape inference if found

  #55410 merged Mar 29, 2022

• Fix training accuracy on AArch64 in FC layers

  #55404 merged Mar 28, 2022

• Refactoring of the Converter for Binary Operations

  #55245 merged Mar 28, 2022

• Removes third_party/mkl_dnn/onednn-acl-bf16.patch

  #55401 merged Mar 28, 2022

• [ROCm] Correcting the output size for the RFFT call in `//tensorflow/compiler/xla/service/gpu/tests:mlir_fft_test`

  #55265 merged Mar 28, 2022

• [PluggableDevice] Add DEVICE_DEFAULT int32 registration for Concat and ConcatV2

  #55385 merged Mar 28, 2022

• [PluggableDevice] Move ZerosLike DEVICE_DEFAULT registration outside of CUDA ifdef

  #55386 merged Mar 28, 2022

• [PluggableDevice] Move Max DEVICE_DEFAULT registration outside of CUDA ifdef

  #55393 merged Mar 28, 2022

• [PluggableDevice] Move Unpack DEVICE_DEFAULT registration outside of CUDA ifdef

  #55387 merged Mar 28, 2022

• [PluggableDevice] Add DEVICE_DEFAULT int32 registration for Slice

  #55384 merged Mar 28, 2022

• [PluggableDevice] Move Min DEVICE_DEFAULT registration outside of CUDA ifdef

  #55392 merged Mar 28, 2022

• [PluggableDevice] Add DEVICE_DEFAULT int32 registration for Pack

  #55381 merged Mar 28, 2022

• [oneDNN] Add env variable to MKL allocator to control mem allocation.

  #54169 merged Mar 28, 2022

• Avoid loss of precision from using reciprocal

  #55310 merged Mar 28, 2022

• [oneDNN] optimized implementation for eigen LeakyRelu

  #53848 merged Mar 27, 2022

• Update oneDNN version for mkl_dnn_acl_compatible.

  #55352 merged Mar 26, 2022

• [INTEL oneDNN] user mode scratchpad implementation for conv op

  #55235 merged Mar 25, 2022

• [oneDNN] Upgrade oneDNN to 2.6-rc

  #55354 merged Mar 25, 2022

21 Pull requests opened by 17 people

• Add TF_AssignRefVariable

  #55379 opened Mar 26, 2022

• [PluggableDevice] Add DEVICE_DEFAULT int32 registration for strided slice ops

  #55382 opened Mar 26, 2022

• Print error message when failed to load external delegate.

  #55383 opened Mar 26, 2022

• tanh float32 over 1.

  #55394 opened Mar 27, 2022

• Grammar Correction

  #55397 opened Mar 28, 2022

• TFLite evaluation utils: Fix `elinux_armhf` build

  #55402 opened Mar 28, 2022

• Increase timeout on join for preemption test

  #55409 opened Mar 28, 2022

• [oneDNN] Enable Softmax with oneDNN library

  #55411 opened Mar 28, 2022

• [TF-TRT] Fix two problems in python tests.

  #55412 opened Mar 28, 2022

• Refactoring tpu_api_dlsym_initializer

  #55413 opened Mar 28, 2022

• Removing the executor file and link jax to use tpu_api_dlsym_initializer

  #55415 opened Mar 28, 2022

• [INTEL MKL] Fix oneDNN Ubuntu partials and Dockerfiles and remove support for Ubuntu 16.04

  #55418 opened Mar 29, 2022

• Add dct_method argument to tf.image.adjust_jpeg_quality

  #55427 opened Mar 29, 2022

• Converter for LogicalNot operation

  #55428 opened Mar 29, 2022

• Shahrokhi3 pr

  #55429 opened Mar 29, 2022

• [TF-TRT] Slice Instance_OP Bug Fix and Cleaning

  #55431 opened Mar 29, 2022

• [ROCm] Rocm bef executable fixes

  #55439 opened Mar 30, 2022

• Create tensorflow_issue_template.yaml

  #55443 opened Mar 30, 2022

• Fix tf.scatter_nd documentation

  #55460 opened Apr 1, 2022

• [hotfix] rename pylintrc

  #55464 opened Apr 1, 2022

• [ROCm] Switch ROCm TF builds to use ROCm 5.1.0

  #55467 opened Apr 1, 2022

51 Issues closed by 30 people

• Why is profiler dir deleted for non-chief nodes in MWMS in keras?

  #49852 closed Apr 1, 2022

• I am using AD9361z7035 breakout board. using ubuntu version linaro 12.06 I want to install tensorflow lite on it.

  #55463 closed Apr 1, 2022

• Not able to cluster Conv1DTranspose layer

  #54832 closed Apr 1, 2022

• `tf.experimental.numpy.array` should have the same behavior as `numpy.array`, but currently crashes

  #55292 closed Apr 1, 2022

• runpath not including //tensorflow/python for _dtensor_device.so

  #55437 closed Apr 1, 2022

• org.tensorflow:tensorflow-lite-support:0.3.1 has missing files

  #55298 closed Apr 1, 2022

• Notebooks deleted when following beginner tutorial

  #55449 closed Apr 1, 2022

• Exception when saving custom RNN model with a constant in the call function when using SavedModel format

  #48213 closed Apr 1, 2022

• Custom gradient tape-based fit() function is slower than Tf2 fit() and has memory leaking

  #48732 closed Mar 31, 2022

• Can't use multiple tensorboard callbacks in TF2.8

  #55224 closed Mar 31, 2022

• Tensorflow 2.8.0 | Error Message: ImportError: SystemError. | Tensorflow package error?

  #55440 closed Mar 31, 2022

• CPU execution of ops after gradient clipping on windows

  #16864 closed Mar 31, 2022

• Depthwise convolution inside Dataset API throws data_format error

  #22145 closed Mar 31, 2022

• Can I insert a node in a graph with graph_editor?

  #20138 closed Mar 31, 2022

• CustomGraphOptimizer doesn't work

  #55451 closed Mar 31, 2022

• tf.estimator.train's incompatibility with distributed training on Cloud ML Engine is not well-documented

  #23017 closed Mar 31, 2022

• libtensorflowlite XNNPACK support beyond f32

  #42759 closed Mar 31, 2022

• Docker containers with Python 3.7

  #25939 closed Mar 31, 2022

• Didn't find op for builtin opcode 'RESIZE_BILINEAR' version '3'

  #49207 closed Mar 31, 2022

• tf.matmul() fails with ragged inputs of shape [batch_size, None, dims] and transpose_b=True

  #55209 closed Mar 30, 2022

• Bazel build error. Warning on retrieving an archive which is not found and then fails.

  #55426 closed Mar 30, 2022

• TFLite allocate tensors fails: (CONCATENATION) failed to prepare

  #55424 closed Mar 30, 2022

• Wrong definition of arg fixed_length in api_docs/python/tf/raw_ops/DecodePaddedRaw

  #55372 closed Mar 30, 2022

• Build fails to make use of bazelisk, if available.

  #55218 closed Mar 30, 2022

• TensorFlow Lite (2.3.0) Converter Quantization Incompatibility for Coral Edge TPU Compiler

  #43275 closed Mar 29, 2022

• TensorBoard Dev doesn't update upon delete operation

  #55407 closed Mar 29, 2022

• Bazel build ERROR with TF2.4 Cuda11.0 Cudnn8 Python3.8 MSVC2019 bazel 3.1.0

  #50414 closed Mar 29, 2022

• Graidents for weights in tf.keras.layers.GRUCell not being computed when using tf.estamtor

  #45931 closed Mar 29, 2022

• AutoGraph could not transform when fine-tuning HuggingFace Pretrained Model

  #55038 closed Mar 29, 2022

• Memory Usage of a model

  #39605 closed Mar 29, 2022

• ValueError: Python inputs incompatible with input_signature

  #55403 closed Mar 29, 2022

• seq2seq.BasicDecoder not working with seq2seq.SampleEmbeddingSampler

  #55223 closed Mar 29, 2022

• Add support for CUDA 11.1 on Windows 10 for the 8.6 compute capability

  #44750 closed Mar 29, 2022

• Random number generation with tf.function in tf 2.5.0

  #49463 closed Mar 29, 2022

• 'Activation' object has no attribute 'kernel' applying regularization to MobileNetV3

  #55376 closed Mar 28, 2022

• Tensorflow doesn't show the Cuda import messages

  #15613 closed Mar 28, 2022

• RuntimeError: module compiled against API version 0xe but this version of numpy is 0xd

  #55377 closed Mar 28, 2022

• Targets in `tensorflow/core/kernels/fuzzing` directory.

  #55183 closed Mar 28, 2022

• Make TensorFlow build with system packages

  #36127 closed Mar 28, 2022

• Example code of tfd.Independent seems to be incorrect.

  #55212 closed Mar 28, 2022

• tensorflow core while do stdthread create

  #55186 closed Mar 28, 2022

• Fail to convert trained model to TensorFlow Lite (integer only and unsigned integer)

  #55198 closed Mar 28, 2022

• tfa.activations.snake - wrong function in docs

  #50737 closed Mar 28, 2022

• TensorFlow Developer Certificate error sign in on pycharm

  #55027 closed Mar 27, 2022

• Request for grouped convolutions on CPU

  #53631 closed Mar 26, 2022

• Can we add a check in Model.fit on dataset element_spec?

  #47618 closed Mar 26, 2022

• Using Numpy BitGenerator with TF-1.15 (?)

  #55299 closed Mar 26, 2022

• tf.control_dependencies fails to update the dependent op.

  #19368 closed Mar 26, 2022

• Does eager execution allow dynamic batching like Tensorflow fold?

  #19478 closed Mar 26, 2022

• Documentation for loading `Estimator` and using it to `predict`

  #48309 closed Mar 26, 2022

• Training is slow when using tf.keras.utils.Sequence with large Numpy arrays

  #44266 closed Mar 26, 2022

32 Issues opened by 30 people

• undefined symbol when linking tf code with custom op

  #55466 opened Apr 1, 2022

• Possibly platform related tflite conversion issue

  #55461 opened Apr 1, 2022

• GRU performance severely degraded inside tf.function with Apple m1 chip

  #55455 opened Mar 31, 2022

• Keras cant load model with tf.where when dtype is float64

  #55454 opened Mar 31, 2022

• TensorRT conversion fails with 5D input data

  #55452 opened Mar 31, 2022

• Python Import error for tensorflow, tflearn and nltk libraries!

  #55450 opened Mar 31, 2022

• X86_64 - TFLite different inference results depending on whether it was compiled with XNNpack delegate or not

  #55448 opened Mar 31, 2022

• How about exporting `Tensor::FromProto` to Python API?

  #55447 opened Mar 31, 2022

• Checkpoint variable names are assigned "variables/0/..." etc if keras model has more than two nested layers/models

  #55446 opened Mar 31, 2022

• RuntimeError: Encountered unresolved custom op: PyFunc.Node number 559 (PyFunc) failed to prepare.

  #55445 opened Mar 31, 2022

• what is the order when I call tf.keras.Layers.BatchNormation.get_weights() ? my tensorflow version is 2.2.0?

  #55444 opened Mar 31, 2022

• Pylint phatom checks

  #55442 opened Mar 30, 2022

• [Saved Model] Loading Saved Model & GraphDef mostly doesn't respect device placement

  #55441 opened Mar 30, 2022

• model is loaded as Loader._recreate_base_user_object.<locals>._UserObject

  #55438 opened Mar 30, 2022

• TF 2.8.0 NNApi createInterpreter fails on Android API 32 devices

  #55436 opened Mar 30, 2022

• Loss names are not equal to the outputs dictionary keys of a model

  #55435 opened Mar 30, 2022

• tensorflow-macos version 2.8.0 cannot import keras

  #55433 opened Mar 29, 2022

• MobileNetV3 reloading weights from version 2.8 gives error but not 2.7

  #55432 opened Mar 29, 2022

• `pfor`/`tf.vectorize_map`: Improve feedback/hints messages when it `fallback_to_while_loop`

  #55425 opened Mar 29, 2022

• Customize TF tensor multiplication

  #55423 opened Mar 29, 2022

• Parameter init_epoch can not be identified when calling model.train()

  #55422 opened Mar 29, 2022

• How can I take CPU variable as input of a custom GPU operation?

  #55421 opened Mar 29, 2022

• RuntimeError: Encountered unresolved custom op: StridedSlice.

  #55419 opened Mar 29, 2022

• How to update tf.Variable inplace using custom op?

  #55417 opened Mar 29, 2022

• Known issues with new RUN_EAGER_OP_AS_FUNCTION feature

  #55414 opened Mar 28, 2022

• tf.nn.ctc_loss is more numerically unstable when providing sparse instead of dense labels

  #55405 opened Mar 28, 2022

• !saved_model_cli run command does not allow placeholder for --dir

  #55400 opened Mar 28, 2022

• TFLite Failed to Allocate Tensor

  #55399 opened Mar 28, 2022

• Calling `model.compile()` multiple times leads to memory leak

  #55398 opened Mar 28, 2022

• tanh(float(7~8)) = 1.0000001 in XLA

  #55390 opened Mar 27, 2022

• TensorflowLite Undefined symbol: _TfLiteTensorCopy

  #55389 opened Mar 27, 2022

• module 'tensorflow_federated.python.learning' has no attribute 'reconstruction'

  #55388 opened Mar 26, 2022

225 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-TRT] Add support for non-frozen models with legacy reference variables (VariableV2)

  #55337 commented on Mar 31, 2022 • 17 new comments

• Default EQUALITY_OPERATORS in Autograph

  #55317 commented on Mar 31, 2022 • 16 new comments

• [TF-TRT] Adding `TF_TRT_EXPERIMENTAL_FEATURES` environment variable controller

  #55332 commented on Apr 1, 2022 • 14 new comments

• [TF:TRT] Add in-memory timing cache registry

  #54650 commented on Apr 1, 2022 • 13 new comments

• Saving Resource Variables without Copy

  #55206 commented on Apr 1, 2022 • 10 new comments

• [Autograph] add min max builtins

  #55363 commented on Mar 28, 2022 • 9 new comments

• Updated log_poisson_loss() function

  #51765 commented on Mar 30, 2022 • 8 new comments

• Add GPU implementation of SparseSliceGrad

  #55055 commented on Mar 31, 2022 • 8 new comments

• [TF-TRT] Add TF-TRT converter for ZerosLike and OnesLike

  #54402 commented on Apr 1, 2022 • 7 new comments

• Tracker for TensorFlow-DirectML in TF 2.9.0

  #55226 commented on Mar 31, 2022 • 6 new comments

• Linking process takes forever

  #54714 commented on Mar 31, 2022 • 6 new comments

• Failed to build libtensorflow (v2.8.0 + bazel 4.2.1) from source on Apple M1

  #55369 commented on Mar 29, 2022 • 5 new comments

• benchmark_model: Could not create Hexagon delegate: platform may not support delegate or required libraries are missing

  #55364 commented on Apr 1, 2022 • 5 new comments

• Creating ragged tensors is incredibly slow

  #47853 commented on Mar 31, 2022 • 4 new comments

• Initial running of TensorFlow on WSL Ubuntu 20.04 looks very very slow.

  #55067 commented on Mar 31, 2022 • 4 new comments

• tf.nn.ctc_loss behaviour changes depending on whether dense or sparse labels are provided

  #55328 commented on Apr 1, 2022 • 4 new comments

• [TFLite] Add INT8 and INT16x8 support to NON_MAX_SUPPRESSION_V4 / V5 operators

  #51785 commented on Apr 1, 2022 • 4 new comments

• Optimized pad kernel

  #54494 commented on Mar 31, 2022 • 4 new comments

• Add a root cause arg to the fallback converter

  #55192 commented on Mar 29, 2022 • 4 new comments

• Test to cover the `import_meta_graph` error on MutableHashTable

  #55200 commented on Apr 1, 2022 • 4 new comments

• Fix invalid examples for `tf.RaggedTensor.__abs__`

  #55222 commented on Mar 31, 2022 • 4 new comments

• Converter for LogicalOr and LogicalAnd operations

  #55275 commented on Apr 1, 2022 • 4 new comments

• Ppc build fix

  #55306 commented on Apr 1, 2022 • 4 new comments

• undefined references on tflite x86_64 cross compiled with NDK Android toolchain

  #55336 commented on Mar 29, 2022 • 3 new comments

• Error while trying to serilize Image Captioning keras model '_UserObject' object is not callable'

  #35451 commented on Mar 30, 2022 • 3 new comments

• Segmentation fault when running optimization step with 3d convolution

  #16139 commented on Mar 30, 2022 • 3 new comments

• TFLITE conversion Failed - Custom Model

  #54155 commented on Mar 30, 2022 • 3 new comments

• [TransformGraph & TF2 SavedModel] - Missing API

  #43688 commented on Mar 30, 2022 • 3 new comments

• Running L-BFGS-B optimizer in TF2

  #48167 commented on Mar 31, 2022 • 3 new comments

• tf.load_op_library doesn't load .so file and throws file not found error although the file exists

  #55375 commented on Apr 1, 2022 • 3 new comments

• Tensorflow build fails with "Multiple matches are not allowed unless one is unambiguously more specialized" on linux_s390x

  #53164 commented on Apr 1, 2022 • 3 new comments

• [INTEL oneDNN] Fix mkl_fused_ops_test failure and disable blocked format

  #54379 commented on Mar 29, 2022 • 3 new comments

• Add default settings for cloud tpu client to support AI Platform

  #54455 commented on Apr 1, 2022 • 3 new comments

• [oneDNN] Fuse conv2d+biasAdd/FusedBatchNorm+sigmoid+mul into fusedConv2D

  #55180 commented on Mar 30, 2022 • 3 new comments

• [oneDNN] Fix to some benchmark tests when GPU is not available

  #55193 commented on Mar 31, 2022 • 3 new comments

• TFX Starter Tutorial Typo

  #55282 commented on Mar 28, 2022 • 2 new comments

• Feature request: Add left_side flag and functionality to tf.linalg.triangular_solve

  #55371 commented on Mar 28, 2022 • 2 new comments

• Build TensorFlow Lite for iOS on macOS Mojave

  #55268 commented on Mar 30, 2022 • 2 new comments

• TF-lite conversion of complex abs layer not working with integer quantization with fallbacks

  #55349 commented on Mar 31, 2022 • 2 new comments

• Failed to use evaluation function from TF, TypeError: tf__loss() after loading a saved model

  #55220 commented on Mar 31, 2022 • 2 new comments

• Redundant `create_padding_mask(inp)` masks

  #55313 commented on Mar 31, 2022 • 2 new comments

• un-deprecate `write_grads` for `fit`

  #31173 commented on Mar 31, 2022 • 2 new comments

• `tf.strided_slice()` option for `inclusive_end`

  #55237 commented on Mar 31, 2022 • 2 new comments

• Error during Inference of LSTM Tflite model

  #55065 commented on Mar 31, 2022 • 2 new comments

• Setting `converter.inference_type=uint8` does not produce quantized TFLite model of uint8 weight data type

  #55135 commented on Apr 1, 2022 • 2 new comments

• Is tf2xla used even when jit_compile=False

  #55207 commented on Apr 1, 2022 • 2 new comments

• Add support for padding and cropping to tf.keras.layers.experimental.preprocessing.Resizing

  #46191 commented on Apr 1, 2022 • 2 new comments

• small discrepancies between images loaded with `tf.keras.preprocessing.image_dataset_from_directory()` and `tf.keras.preprocessing.image.load_img()`

  #46255 commented on Apr 1, 2022 • 2 new comments

• Feature/device-annotation

  #54510 commented on Apr 1, 2022 • 2 new comments

• [ROCm] fix triangular solve

  #54771 commented on Mar 31, 2022 • 2 new comments

• Simplify the DoPoolForward/Backward

  #55281 commented on Apr 1, 2022 • 2 new comments

• Refactoring of the converter for Activation operators.

  #55319 commented on Apr 1, 2022 • 2 new comments

• tf.cond within vectorized_map results in unknown output shape

  #46818 commented on Mar 25, 2022 • 1 new comment

• [bug] gradients of scatter_nd_add return None

  #12473 commented on Mar 25, 2022 • 1 new comment

• Error message of scatter_update is misleading

  #12843 commented on Mar 25, 2022 • 1 new comment

• How to fix 'Tidx' type in TensorFlowOpLayer (Inconsistent values for attr 'Tidx' DT_FLOAT vs. DT_INT32 while building NodeDef)

  #47309 commented on Mar 26, 2022 • 1 new comment

• Feature Request - Deeplab TFLITE Android Application

  #23747 commented on Mar 26, 2022 • 1 new comment

• Error in converting tensorflow model hdf5 to onnx format

  #52989 commented on Mar 26, 2022 • 1 new comment

• windows/subprocess.cc should log the command line for a failing invocations

  #48089 commented on Mar 27, 2022 • 1 new comment

• Failed to build TensorFlow package builder

  #50340 commented on Mar 27, 2022 • 1 new comment

• [Accuracy] round issue on tf.nn.conv2d, it does not return the same results for a kernel depth size < 8 vs kernel depth size > 16

  #55326 commented on Mar 27, 2022 • 1 new comment

• Blas xGEMV launch failed error when running on GPU.

  #52222 commented on Mar 28, 2022 • 1 new comment

• how to change tensorflow version if build from source?

  #55297 commented on Mar 28, 2022 • 1 new comment

• There are some error when convert Mask RCNN model to .tflite

  #25415 commented on Mar 28, 2022 • 1 new comment

• The tensorflow training freezes for no reason

  #55283 commented on Mar 28, 2022 • 1 new comment

• Tensorflow Model Compiler Metrics Error: NaN Outputs

  #55277 commented on Mar 28, 2022 • 1 new comment

• The code says that it failed to save and then never ran properly again. basically i upload a picture and then it shows the right prediction but the second picture it run into error but if i start with second photo as my first inout then it show the correct prediction

  #55289 commented on Mar 28, 2022 • 1 new comment

• `tf.experimental.numpy.floor_divide` is inconsistent with numpy version when divide by zero

  #54856 commented on Mar 28, 2022 • 1 new comment

• Failed to call tf.convert_to_tensor for a tf.RaggedTensor with a PlaceHolder value

  #55373 commented on Mar 28, 2022 • 1 new comment

• Auto comment generation project - model training loop crash

  #55324 commented on Mar 28, 2022 • 1 new comment

• After finetuning a BERT model max_position_embeddings=128, my saved model is not reflecting the smaller sequence length

  #54687 commented on Mar 28, 2022 • 1 new comment

• Issue with XLA devices or MLIR optimization?

  #55014 commented on Mar 28, 2022 • 1 new comment

• TFHub Object Detection Models Cannot Quantize

  #55073 commented on Mar 28, 2022 • 1 new comment

• tf.gradient returns None

  #55262 commented on Mar 28, 2022 • 1 new comment

• CUDA_ERROR_ILLEGAL_ADDRESS when calling tf.debugging.assert_less using tf.uint8

  #53243 commented on Mar 28, 2022 • 1 new comment

• Stateful ConvLSTM2D reset_states() throws error

  #48863 commented on Mar 28, 2022 • 1 new comment

• Misleading behavior on matrix and vector division, which is in contradiction with the documentation in nn.softmax

  #53112 commented on Mar 29, 2022 • 1 new comment

• Allow cloning of tf.keras.Model subclass

  #47048 commented on Mar 29, 2022 • 1 new comment

• Typo in https://blog.tensorflow.org/2022/02/boost-your-models-accuracy.html?linkId=8033010

  #54406 commented on Mar 29, 2022 • 1 new comment

• Plugin implementation of AssignVarOp

  #55295 commented on Mar 29, 2022 • 1 new comment

• Model with both complex and real weights does not work with tf.distribute.MirroredStrategy

  #53275 commented on Mar 29, 2022 • 1 new comment

• Can I use tensorflow on Windows10 with c++ and gpu support

  #8773 commented on Mar 29, 2022 • 1 new comment

• Java: add support for Alpine Linux (Docker)

  #27643 commented on Mar 29, 2022 • 1 new comment

• XLA reports error with 1000 steps of static_bidirectional_rnn

  #14507 commented on Mar 29, 2022 • 1 new comment

• PyCharm doesn't resolve anything under tensorflow.keras

  #53144 commented on Mar 29, 2022 • 1 new comment

• Large multinomial sampling on GPU causes OOM

  #18058 commented on Mar 29, 2022 • 1 new comment

• Binary add op BF16 has lower performance than FP32

  #37975 commented on Mar 29, 2022 • 1 new comment

• CUPTI events are missing from tf.train.Server

  #18849 commented on Mar 29, 2022 • 1 new comment

• Feature Request: Sparse Cholesky decomposition

  #15910 commented on Mar 29, 2022 • 1 new comment

• Question: Can we get all the intermidiate tensors when using gpu(opencl) delegate??

  #51126 commented on Mar 29, 2022 • 1 new comment

• Feature request: tf.Print should either print (not log), or accept a log level

  #16523 commented on Mar 29, 2022 • 1 new comment

• Extremely slow grads computation in dilatied convlution with a large dilation rate

  #54279 commented on Mar 29, 2022 • 1 new comment

• Array conv_23/Conv_1/BatchNorm/FusedBatchNorm, which is an input to the Conv operator producing the output array conv_3//0/Conv/Relu6, is lacking min/max data, which is necessary for quantization.

  #27952 commented on Mar 29, 2022 • 1 new comment

• Build Fails with TF 2.6 on VS2019 (Win 10)

  #51573 commented on Mar 30, 2022 • 1 new comment

• C++ compile tensorflow

  #50744 commented on Mar 30, 2022 • 1 new comment

• KerasClassifier, GridSearchCV ignore with tf.device('cpu:0')

  #47085 commented on Mar 30, 2022 • 1 new comment

• Hexagon delegate doesn`t work on Android devices, cannot be loaded library UnsatisfiedLinkError (CMake)

  #55269 commented on Mar 30, 2022 • 1 new comment

• How to save tf.data.Dataset object?

  #42934 commented on Mar 30, 2022 • 1 new comment

• CVE-2022-23593 and CVE-2022-23592 are reported incorrectly for TensorFlow 2.7.1

  #55132 commented on Mar 30, 2022 • 1 new comment

• TFLite with Hexagon delegate produces wrong results for a particular model

  #54481 commented on Mar 30, 2022 • 1 new comment

• AttributeError: 'arguments' object has no attribute 'posonlyargs'

  #55325 commented on Mar 30, 2022 • 1 new comment

• AutoGraph saved model with uint8 input will not convert to tflite

  #40268 commented on Mar 30, 2022 • 1 new comment

• tf concatination layers compatible with tfp

  #44706 commented on Mar 30, 2022 • 1 new comment

• I cannot start my tensorflow certificate exam due to error

  #55128 commented on Mar 30, 2022 • 1 new comment

• Will Lamba.compute_output_shape change "inbound_nodes" of Lambda layer?

  #47977 commented on Mar 30, 2022 • 1 new comment

• keras.Model.save_weights is overwriting all_model_checkpoint_paths

  #41553 commented on Mar 30, 2022 • 1 new comment

• tensorflow.keras.utils.Sequence does not work with TPU

  #47769 commented on Mar 30, 2022 • 1 new comment

• Boolean_mask can't be compiled by XLA on CPU

  #44904 commented on Mar 30, 2022 • 1 new comment

• TPU compile fail using the new 2.4.0 version

  #45852 commented on Mar 30, 2022 • 1 new comment

• version 2.3.0 build error: user_ops_gen_cc: undefined symbol

  #42221 commented on Mar 31, 2022 • 1 new comment

• OSError: [Errno 9] Bad file descriptor raised on program exit

  #50487 commented on Mar 31, 2022 • 1 new comment

• A/libc: Fatal signal 11 (SIGSEGV), code 2 (SEGV_ACCERR), fault addr 0xb9124f40 in tid 9093 (superresolution), pid 9093 (superresolution)

  #55123 commented on Mar 31, 2022 • 1 new comment

• list of operators for which you will need custom implementations: StatelessWhile, TensorListFromTensor, TensorListReserve, TensorListStack, While.

  #37134 commented on Mar 31, 2022 • 1 new comment

• TF.data.dataset.cache(path) still uses memory despite the cache file path is given in tf2.8

  #55355 commented on Mar 31, 2022 • 1 new comment

• Feature: Add reduce_average (weighted reduce_mean)

  #7422 commented on Mar 31, 2022 • 1 new comment

• @tensorflow/tfjs: Unable to load model

  #55357 commented on Mar 31, 2022 • 1 new comment

• [PluggableDevice] C-API Symbols not exposed

  #52077 commented on Mar 31, 2022 • 1 new comment

• Build from Source Instructions for Tensorflow C++ on Linux

  #55279 commented on Mar 31, 2022 • 1 new comment

• Using libtensorflow-core.a to enable training on Android

  #31829 commented on Mar 31, 2022 • 1 new comment

• MLIR-based convertor fails to convert Universal Sentence Encoder model

  #34763 commented on Mar 31, 2022 • 1 new comment

• Depth_to_space

  #51421 commented on Mar 31, 2022 • 1 new comment

• quantization deeplabv3(mobielentv2) error

  #20867 commented on Mar 31, 2022 • 1 new comment

• Can't link with TF Lite C API static lib - undefined reference to `tflite::DefaultErrorReporter()'

  #55344 commented on Mar 31, 2022 • 1 new comment

• [bazel] Unrecoverable error while evaluating node '//tensorflow/cc:ops/io_ops_gen_cc

  #22459 commented on Mar 31, 2022 • 1 new comment

• Wav to Spectogram wrong expected height

  #22838 commented on Mar 31, 2022 • 1 new comment

• compute_mask not called when custom layer does not alter inputs and mask is not None

  #32322 commented on Mar 31, 2022 • 1 new comment

• Factor limit of `tf.image.adjust_contrast` and `tf.image.adjust_saturation`?

  #55374 commented on Mar 31, 2022 • 1 new comment

• Segmentation Fault (core dumped) Issue

  #54747 commented on Mar 31, 2022 • 1 new comment

• [v2.8.0]: Cross compiling libtensorflow.so for ARM64 fails

  #54353 commented on Mar 31, 2022 • 1 new comment

• tf.estimator.BoostedTreesEstimator returns all zero predictions when using custom loss_fn involving linear loss (zero second derivative)

  #48119 commented on Mar 31, 2022 • 1 new comment

• Encounter Abort(core dump) in ClipOpsTest when changing shapes

  #55201 commented on Mar 31, 2022 • 1 new comment

• backing_device changes depending on the tensor's dtype

  #46833 commented on Mar 31, 2022 • 1 new comment

• InvalidArgumentError: Node '.../while_grad': Connecting to invalid output 4 of source node while which has 4 outputs

  #54458 commented on Mar 31, 2022 • 1 new comment

• Per thread allocation of MKL primitives

  #54979 commented on Mar 31, 2022 • 1 new comment

• [TFLite] Failed to build iOS framework: thread-local storage is not supported for the current target

  #55322 commented on Mar 31, 2022 • 1 new comment

• [RNN] Max and min for dynamic tensors should be recorded during calibration: Failed for tensor arg1 Empty min/max for tensor arg1

  #55267 commented on Mar 31, 2022 • 1 new comment

• About how does buffer size from Dataset.shuffle() influence Dataset.cache()

  #55361 commented on Apr 1, 2022 • 1 new comment

• MirroredStrategy was not working.

  #55358 commented on Apr 1, 2022 • 1 new comment

• DeeplabV3 custom dataset, inference problem black images

  #55360 commented on Apr 1, 2022 • 1 new comment

• tf.repeat fails with jit_compile when repeats is a tensor!

  #52905 commented on Apr 1, 2022 • 1 new comment

• DOC: explain the meaning of model.fit() history tracking in a distributed scenario

  #55366 commented on Apr 1, 2022 • 1 new comment

• TensorFlow android throw java.lang.OutOfMemoryError when building project, but there's free 20GB memory.

  #55368 commented on Apr 1, 2022 • 1 new comment

• How to convert Keras LSTM model into tf.nn.rnn_cell.LSTMCell model

  #55345 commented on Apr 1, 2022 • 1 new comment

• [FEATURE REQUEST] tf.scatter_nd doesn't support half types

  #20219 commented on Apr 1, 2022 • 1 new comment

• Performance regression: Tensorflow 2.5.0

  #51145 commented on Apr 1, 2022 • 1 new comment

• Regression: TFLiteConverter changes output order again since 2.7

  #55242 commented on Apr 1, 2022 • 1 new comment

• Unrecognized term used by the documentation of the class "tensorflow.compat.v1.variable_scope".

  #54449 commented on Apr 1, 2022 • 1 new comment

• [TF2.0] Change default types globally

  #26033 commented on Apr 1, 2022 • 1 new comment

• wierd behavior while running the inference on 2 different deep learning models(pb graphs) at the same

  #53834 commented on Apr 1, 2022 • 1 new comment

• Added an argument to control the padding for flatten_atrous

  #28049 commented on Mar 29, 2022 • 1 new comment

• Fix discrepancy between tf.io.gfile.mkdir and os.mkdir's created mode

  #33312 commented on Mar 29, 2022 • 1 new comment

• Try RecursivelyCreateDir Except os.makedirs solution to Create Missing Path

  #41893 commented on Mar 29, 2022 • 1 new comment

• C++ canonicalize_function_inputs WIP

  #43053 commented on Mar 29, 2022 • 1 new comment

• [Intel MKL] Fuse group of primitive ops for batch normalization to Fu…

  #44950 commented on Mar 29, 2022 • 1 new comment

• [INTEL MKL] Convert necessary PlaceholderWithDefault or Placeholder to Constant f…

  #45822 commented on Mar 29, 2022 • 1 new comment

• Fix #45839: tf.distribute.CollectiveAllReduceStrategy can't load model from checkpoint

  #45898 commented on Mar 29, 2022 • 1 new comment

• [Intel MKL] Fix shape inference for Quantized{Conv2D/DepthwiseConv2D}-like operations

  #46779 commented on Mar 29, 2022 • 1 new comment

• [GO] Memory leak when loading in a string source in encodeTensorWithSlices

  #48051 commented on Mar 29, 2022 • 1 new comment

• [INTEL MKL] - Adding 'oneDNN' partials and Dockerfiles for 'RedHat 8'

  #48056 commented on Mar 29, 2022 • 1 new comment

• Cleanup for more consistent use of c_api_macros.h, throughout tensorflow C APIs.

  #48084 commented on Mar 29, 2022 • 1 new comment

• Add elastic training for workers in PSv2

  #48601 commented on Mar 29, 2022 • 1 new comment

• Numpy and dataframe converter

  #49278 commented on Mar 29, 2022 • 1 new comment

• [TFLite] Quantized int8/16 support for TILE operator

  #50424 commented on Mar 29, 2022 • 1 new comment

• [Go] add bazel rules

  #50934 commented on Mar 29, 2022 • 1 new comment

• [TFLite] Add int8 and int16x8 support for CONV_3D operator

  #51527 commented on Mar 29, 2022 • 1 new comment

• [INTEL MKL] BF16/FP32 RandomUniform vectorization opt on CPU

  #51544 commented on Mar 29, 2022 • 1 new comment

• [TFLite] Add int8 and int16x8 support for TRANSPOSE_CONV_3D operator

  #51572 commented on Mar 29, 2022 • 1 new comment

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

  #51599 commented on Mar 30, 2022 • 1 new comment

• [PluggableDevice] Add TF_IsHostMemoryInput and TF_IsHostMemoryOutput to the C API

  #51759 commented on Mar 30, 2022 • 1 new comment

• [BUG FIX]fix mkl relu6 backward bug.

  #52002 commented on Mar 30, 2022 • 1 new comment

• [TFLite] Use integer-only operations for the quantized reference REDUCE_MEAN and REDUCE_SUM kernels

  #52014 commented on Mar 30, 2022 • 1 new comment

• [TF:TRT] bilinear resize refactor + workaround

  #52188 commented on Mar 30, 2022 • 1 new comment

• Add mlir-hlo unfuse-batch-norm-training pattern

  #52241 commented on Mar 31, 2022 • 1 new comment

• [TF:TRT] Use the third_party TRT-OSS plugin for CombinedNMS

  #53414 commented on Mar 30, 2022 • 1 new comment

• [TF:TRT] Add Grappler optimizer that executes rewrites for TF-TRT conversion preperation

  #53659 commented on Mar 26, 2022 • 1 new comment

• TensorFlow Lite Evaluation Tools CMake Compilation

  #53691 commented on Mar 30, 2022 • 1 new comment

• TensorFlow Lite CMake dependencies versions automatically synced with Bazel configuration

  #54220 commented on Mar 30, 2022 • 1 new comment

• Add appropriate value check for `tf.clip_by_norm`

  #54430 commented on Mar 30, 2022 • 1 new comment

• [TF:TRT] Enable out-of-tree op converter registration

  #54905 commented on Mar 28, 2022 • 1 new comment

• error executing command crosstool_wrapper_driver_is_not_gcc

  #53680 commented on Mar 25, 2022 • 0 new comments

• dataset as_numpy_iterator() failing when using RaggedTensor

  #53149 commented on Mar 28, 2022 • 0 new comments

• `tf.random.shuffle` is deterministic when XLA is used

  #52277 commented on Mar 29, 2022 • 0 new comments

• @tf.function triggers "AttributeError: 'Tensor' object has no attribute 'ndim'"

  #48612 commented on Mar 29, 2022 • 0 new comments

• Segmentation fault when passing tf.string tensor arguments in TFLite

  #53111 commented on Mar 29, 2022 • 0 new comments

• C-API Binaries for 32-bit for x86 instruction set

  #55257 commented on Mar 29, 2022 • 0 new comments

• SeparableConv2D from tf.keras.layers and tf.layers gives different results

  #18338 commented on Mar 29, 2022 • 0 new comments

• Gather/Slice/StridedSlice Gradients Support in the C++ API

  #9645 commented on Mar 29, 2022 • 0 new comments

• xla compile op cache many tensors to out of memory .

  #52645 commented on Mar 29, 2022 • 0 new comments

• tf.device scope not working correctly

  #44510 commented on Mar 29, 2022 • 0 new comments

• Changing self properties does not trigger retracing

  #44980 commented on Mar 31, 2022 • 0 new comments

• `tf.ragged.row_splits_to_segment_ids` lack input validation

  #54849 commented on Mar 31, 2022 • 0 new comments

• c++ use LoadSavedModel can't choose gpu?

  #53187 commented on Mar 31, 2022 • 0 new comments

• Tensor evaluating to None

  #55184 commented on Mar 31, 2022 • 0 new comments

• Code under tf.init_scope behaves differently then code manually lifted from graph.

  #55303 commented on Mar 31, 2022 • 0 new comments

• -D_GLIBCXX_USE_CXX11_ABI=1 increases a lot RAM usage

  #35160 commented on Apr 1, 2022 • 0 new comments

• the shared network inside a compiled keras model got modified after resetting the trainable attribute of the shared network

  #41796 commented on Apr 1, 2022 • 0 new comments

• tf.nn.sigmoid_cross_entropy_with_logits should support broadcasting

  #51967 commented on Apr 1, 2022 • 0 new comments

• Segmentation fault for tflite Interpreter

  #47996 commented on Apr 1, 2022 • 0 new comments

• tf.sparse.softmax lack support for float16

  #53657 commented on Apr 1, 2022 • 0 new comments

• tensorflow/lite/schema/upgrade_schema.py has a docstring arg out-of-line

  #53566 commented on Apr 1, 2022 • 0 new comments

• Module structure not recognized by Visual Studio Code Linter

  #54435 commented on Apr 1, 2022 • 0 new comments

• Remove duplicated Env::Default to fix gcs issue on Windows

  #49520 commented on Mar 30, 2022 • 0 new comments

• Vscode devcontainer

  #51439 commented on Mar 27, 2022 • 0 new comments

• [PluggableDevice] Add `TF_OpKernelConstruction_GetNodeDef`

  #52157 commented on Mar 30, 2022 • 0 new comments

• fix moments float16 cast bug

  #52217 commented on Apr 1, 2022 • 0 new comments

• [tensorflow/compiler/jit/increase_dynamism_for_auto_jit_pass.cc] Use correct size type for `NumElements` iterator

  #52464 commented on Mar 30, 2022 • 0 new comments

• Update `custom_gradient.py`: fix comment w.r.t `gradients` now being `GradientTape`

  #53005 commented on Mar 30, 2022 • 0 new comments

• Update autoclustering_xla.ipynb

  #53012 commented on Mar 30, 2022 • 0 new comments

• Adding support for Zstd Compression

  #53385 commented on Apr 1, 2022 • 0 new comments

• [Go] Improve tests: fix test issues, add cases and locate assets

  #53469 commented on Mar 30, 2022 • 0 new comments

• [Go] Document use of tensorflow.ConfigProto protocol message to set session options

  #53521 commented on Mar 30, 2022 • 0 new comments

• Update post_training_quantization.md

  #53574 commented on Mar 30, 2022 • 0 new comments

• [tensorflow/lite/schema/upgrade_schema.py] Move `(see :schema.fbs).` up to `operator_type` arg descriptor

  #53576 commented on Mar 30, 2022 • 0 new comments

• Update tflite::gpu::Delegate in delegate.cc

  #53734 commented on Apr 1, 2022 • 0 new comments

• Add float16 support for tf.sparse.softmax

  #53741 commented on Mar 30, 2022 • 0 new comments

• Add encoding support for make_csv_dataset and File IO

  #53838 commented on Mar 29, 2022 • 0 new comments

• TensorFlow Lite kernel tests CMake build fix - duplicated symbols

  #54253 commented on Mar 29, 2022 • 0 new comments

• Add complex support for `tf.sparse.segment_sum`

  #54357 commented on Mar 29, 2022 • 0 new comments

• Update sonartype Maven repo URL to HTTPS

  #54425 commented on Mar 30, 2022 • 0 new comments

• Fix mhlo.concatenate pattern when operand has unknown rank

  #54491 commented on Mar 29, 2022 • 0 new comments

• Adding executable code for `tf.sparse.softmax`

  #55051 commented on Mar 27, 2022 • 0 new comments

• [TF:TRT] INT8 calibration in dynamic shape mode

  #55166 commented on Apr 1, 2022 • 0 new comments

• Fix invalid input for `tf.gather_nd` with `batch_dims`

  #55210 commented on Mar 29, 2022 • 0 new comments

• [tflite] replace non-ASCII single quote

  #55264 commented on Mar 30, 2022 • 0 new comments

• Integrating cuBLASLt into XLA

  #55351 commented on Apr 1, 2022 • 0 new comments