tensorflow / tensorflow

107 Pull requests merged by 34 people

• XLA: Print nicer message when trying to traverse HLO computation

  #43458 merged Sep 23, 2020

• Revert "Revert "Add symmetric int16 support to tflu softmax reference kernel""

  #43320 merged Sep 22, 2020

• flatbuffers changes for SHAPE op in TFLM

  #43384 merged Sep 22, 2020

• for #43320 Changed gen_lut params from std::function to function pointer

  #43434 merged Sep 22, 2020

• GPU-deterministic tf.image.resize (bilinear)

  #39243 merged Sep 22, 2020

• Document possible values of the nnapi_execution_preference option

  #43296 merged Sep 22, 2020

• Rename horizontal_fusion to horizontal_loop_fusion.

  #43316 merged Sep 22, 2020

• Update version numbers for TensorFlow 2.3.1

  #43446 merged Sep 22, 2020

• Update version numbers for TensorFlow 2.2.1

  #43445 merged Sep 22, 2020

• Update version numbers for TensorFlow 2.1.2

  #43444 merged Sep 22, 2020

• Update version numbers for TensorFlow 2.0.3

  #43443 merged Sep 22, 2020

• Update version numbers for TensorFlow 1.15.4

  #43442 merged Sep 22, 2020

• Update release notes for TensorFlow 2.3.1

  #43441 merged Sep 22, 2020

• Update release notes for TensorFlow 2.2.1

  #43440 merged Sep 22, 2020

• Update release notes for TensorFlow 2.1.2

  #43439 merged Sep 22, 2020

• Update release notes for TensorFlow 2.0.3

  #43438 merged Sep 22, 2020

• Update release notes for TensorFlow 1.15.4

  #43437 merged Sep 22, 2020

• Functional preprocessing stage

  #42019 merged Sep 21, 2020

• [tflite] make `//tensorflow/lite/kernels:all` build

  #42365 merged Sep 21, 2020

• Fix missing import

  #43414 merged Sep 21, 2020

• Fix tests broken during cherry-picks

  #43413 merged Sep 21, 2020

• Fix tests broken by merges

  #43412 merged Sep 21, 2020

• Fix tests broken by merge conflicts

  #43411 merged Sep 21, 2020

• Fix broken tests after cherrypicks

  #43410 merged Sep 21, 2020

• Remove import that is not needed

  #43408 merged Sep 21, 2020

• Remove import that is not needed

  #43407 merged Sep 21, 2020

• Remove import that is not needed

  #43406 merged Sep 21, 2020

• Fix import path

  #43405 merged Sep 21, 2020

• Fix import path

  #43404 merged Sep 21, 2020

• Fix import path

  #43403 merged Sep 21, 2020

• Cast away a const in intermediate API

  #43402 merged Sep 20, 2020

• Cast away a const in intermediate API

  #43401 merged Sep 20, 2020

• Fix typo in macro

  #43400 merged Sep 20, 2020

• Fix typo in macro

  #43399 merged Sep 20, 2020

• Fix typo in macro

  #43398 merged Sep 20, 2020

• Fix import path

  #43396 merged Sep 20, 2020

• Fix import path

  #43395 merged Sep 20, 2020

• Fix import path

  #43394 merged Sep 20, 2020

• Solve leftover from merge conflict

  #43391 merged Sep 20, 2020

• Solve leftover from merge conflict

  #43392 merged Sep 20, 2020

• Solve leftover from merge conflict

  #43390 merged Sep 20, 2020

• Solve leftover from merge conflict

  #43389 merged Sep 20, 2020

• Solve leftover from merge conflict

  #43388 merged Sep 20, 2020

• Patch for TF 2.3.1

  #43358 merged Sep 20, 2020

• Patch for TF 2.2.1

  #43357 merged Sep 20, 2020

• Patch for TF 2.1.2

  #43356 merged Sep 20, 2020

• Patch for TF 2.0.3

  #43355 merged Sep 20, 2020

• Patch for TF 1.15.4

  #43354 merged Sep 20, 2020

• Bump sqlite to 3.33.0

  #43346 merged Sep 18, 2020

• Cherry-pick sqlite version bump

  #40733 merged Sep 18, 2020

• Cherry-pick sqlite version bump

  #40732 merged Sep 18, 2020

• Cherry-pick sqlite version bump

  #40731 merged Sep 18, 2020

• Cherry-pick sqlite version bump

  #40730 merged Sep 18, 2020

• Remove scipy dependency.

  #41867 merged Sep 18, 2020

• Remove scipy dependency.

  #41866 merged Sep 18, 2020

• Remove scipy dependency.

  #41865 merged Sep 18, 2020

• Revert "2.3.0-rc2 cherry-pick request: Cherry pick library threadsafestatus "

  #43338 merged Sep 18, 2020

• [CherryPick:r2.3] Fix deprecated usage of collections ABC

  #41355 merged Sep 18, 2020

• 2.3.0-rc2 cherry-pick request: Cherry pick library threadsafestatus

  #41669 merged Sep 18, 2020

• Implementation of a missing hook in XlaDeviceContext

  #43134 merged Sep 18, 2020

• Run SWIG in the default environment

  #42362 merged Sep 18, 2020

• [ROCm] Switch r2.1 to use ROCm 3.7

  #40060 merged Sep 18, 2020

• TFLite library fix for r2.1 branch

  #39773 merged Sep 18, 2020

• Grammar fix

  #43267 merged Sep 18, 2020

• Sagemaker resolver

  #43251 merged Sep 18, 2020

• Update delegates.md with buildin_ops.h header

  #43266 merged Sep 18, 2020

• Also fix single pip Windows GPU renaming on python 3.8

  #43329 merged Sep 18, 2020

• Also fix single pip Windows GPU renaming on python 3.8

  #43328 merged Sep 18, 2020

• Fix rename of gpu pips for single pip package on Windows GPU

  #43327 merged Sep 18, 2020

• Fix rename of gpu pips for single pip package on Windows GPU

  #43326 merged Sep 18, 2020

• Fix rename of gpu pips for single pip package on Windows GPU

  #43325 merged Sep 18, 2020

• Fix #41630: include max_seq_length in cudnn descriptor cache key

  #42634 merged Sep 18, 2020

• Fix 1.15 build and run issues on iOS

  #42902 merged Sep 18, 2020

• Add -D<tag> for every tag specified with TAGS=<tag_list>

  #43314 merged Sep 18, 2020

• Disable a flaky test on mac py38

  #43321 merged Sep 17, 2020

• Grad check v1

  #42344 merged Sep 17, 2020

• [TFLite] Update the operators versions supported by the reference kernels in BuiltinRefOpResolver and add some missing operators

  #43279 merged Sep 17, 2020

• Pin numpy on windows too

  #43318 merged Sep 17, 2020

• Disable more tests on mac

  #43317 merged Sep 17, 2020

• Pin estimator to the right version

  #43315 merged Sep 17, 2020

• Add rel/ pool

  #43313 merged Sep 17, 2020

• [INTEL MKL] Add MKL Conv + Bias + LeakyRelu Fusion

  #42856 merged Sep 17, 2020

• Remove stray backticks in `tf.keras.initializers.Initializer` docstring.

  #43136 merged Sep 17, 2020

• Fix XNNPack delegate type-check for Conv2D bias.

  #43265 merged Sep 17, 2020

• Pin Estimator nightly to latest version used during release

  #43310 merged Sep 17, 2020

• Pin Estimator-nightly to latest version used for 2.2 release

  #43308 merged Sep 17, 2020

• Pin estimator nightly to the latest 2.1 version

  #43306 merged Sep 17, 2020

• Disable test which fails on mac pip

  #43305 merged Sep 17, 2020

• Fix the compilation problem of S3 test cases

  #43222 merged Sep 17, 2020

• Extend documentation for the minimal tflite example.

  #42667 merged Sep 17, 2020

• Fix GatherV2 shape inference

  #42549 merged Sep 17, 2020

• Removed softmax Int16 dependency on std::vector

  #43180 merged Sep 17, 2020

• Fix sanity build

  #43287 merged Sep 17, 2020

• Disable test that fails due to breaking change in absl_py

  #43286 merged Sep 17, 2020

• Update common_win.bat

  #43285 merged Sep 17, 2020

• Fix build on r2.1 branch

  #43284 merged Sep 17, 2020

• Fix missing comma

  #43283 merged Sep 17, 2020

• Fixing build files for broken test/builds

  #43281 merged Sep 16, 2020

• tensor_diag_part doc fix

  #42154 merged Sep 16, 2020

• [Grappler] Fold broadcast select

  #43217 merged Sep 16, 2020

• Fix the compilation problem of HDFS test cases

  #43221 merged Sep 16, 2020

• Implement horizontal input fusion.

  #43051 merged Sep 16, 2020

• TFLu: Support operator only input tensors.

  #42697 merged Sep 16, 2020

• Add missing initialization of filter_offset in fully connected CMSIS-NN kernel

  #43110 merged Sep 16, 2020

• updated documentation for tensorflow lite micro-speech example on NXP…

  #43241 merged Sep 16, 2020

• Add test ReadWhileOverwriting

  #43220 merged Sep 16, 2020

• Do not symlink system protobuf headers but only the required .proto files

  #43153 merged Sep 16, 2020

28 Pull requests opened by 22 people

• TypeError on line 88 should have a condition

  #43260 opened Sep 16, 2020

• TF-TRT Expose TRTEngineInstance class

  #43268 opened Sep 16, 2020

• Mutablehashtable lookup support full size dynamic default values.

  #43269 opened Sep 16, 2020

• [INTEL MKL] MKL DNN 0.x code cleanup - RequantizePerChannel op

  #43274 opened Sep 16, 2020

• [INTEL MKL] MKL DNN 0.x code cleanup - Reshape op

  #43280 opened Sep 16, 2020

• [INTEL MKL] MKL DNN 0.x code clean - input cuponversion op

  #43282 opened Sep 16, 2020

• Various Google C++ style guide fixes

  #43290 opened Sep 17, 2020

• Add GPU kernels for ApplyProximalAdagrad and SparseApplyFtrl

  #43299 opened Sep 17, 2020

• Fixing bug where undocumented FileNotFound error is thrown when trying to use tf.io.gfile.glob()

  #43333 opened Sep 18, 2020

• Modifying get_optimizer_experimental_options

  #43342 opened Sep 18, 2020

• Initialize POD types

  #43343 opened Sep 18, 2020

• Pad() gradient

  #43353 opened Sep 18, 2020

• Define __slots__ for object wrappers

  #43365 opened Sep 19, 2020

• Add Function.experimental_get_tracing_count fixing #37323

  #43366 opened Sep 19, 2020

• Add shape op and tests to TFLM

  #43381 opened Sep 20, 2020

• [TFLM] Added handling of tensors used as outputs for ops but then never used

  #43382 opened Sep 20, 2020

• Add TF_UpdateEdge C API.

  #43383 opened Sep 20, 2020

• add support for mips64 platform

  #43415 opened Sep 21, 2020

• Fixes #42872: map_to_outputs_names always returns a copy

  #43417 opened Sep 21, 2020

• Fixes #43200 - TensorBoard histogram breaks on unsupported weight dtypes

  #43418 opened Sep 21, 2020

• Update to include filesystem:filesystem_interface in libtensorflow_framework.so

  #43424 opened Sep 21, 2020

• Support RESIZE_BILINEAR in TFLu

  #43426 opened Sep 21, 2020

• flatbuffer_conversions changes for resize bilinear

  #43427 opened Sep 21, 2020

• [MLIR] Add involution based folding to TF

  #43428 opened Sep 21, 2020

• Minor spelling tweaks

  #43432 opened Sep 21, 2020

• [INTEL MKL] MKL DNN 0.x code cleanup - Concat op

  #43435 opened Sep 21, 2020

• Do not create unnecessary intermediate lists in tf.nest

  #43454 opened Sep 22, 2020

• Fix casting condition in Embedding layer

  #43466 opened Sep 22, 2020

119 Issues closed by 61 people

• tf.autodiff.ForwardAccumulator fails for Embedding layer

  #43449 closed Sep 23, 2020

• tf.saved_model.save very slow with second-order tf.autodiff.ForwardAccumulator

  #43252 closed Sep 22, 2020

• savermodel quantization arrive TFlite model,on the windows+python Test times are slow

  #42954 closed Sep 22, 2020

• Tflite ops: BatchMatMul, TensorListReserve, TensorListSetItem, TensorListStack

  #42937 closed Sep 22, 2020

• TF 1.15+ mutating sys.modules

  #42930 closed Sep 22, 2020

• Tensorflow developer certificate

  #43462 closed Sep 22, 2020

• Faild build windows

  #43345 closed Sep 22, 2020

• TFLite C API problem inferencing

  #43041 closed Sep 22, 2020

• Unexpected accuracy output from tf.keras.Model.evaluate called on a saved and loaded model

  #43457 closed Sep 22, 2020

• Error while trying to save the decoder of the Image-Captioning model into TFLite file.

  #43292 closed Sep 22, 2020

• [RNN] Error while converting the Decoder in image-Captioning model to TFLite file.

  #43375 closed Sep 22, 2020

• failed precondition: error while reading resource variable block8_sepconv3_bn/gamma

  #43452 closed Sep 22, 2020

• InaccessibleTensorError: The tensor cannot be accessed here: it is defined in another function or code block.

  #43453 closed Sep 22, 2020

• AttributeError: module 'tensorflow' has no attribute 'Dimension'

  #43031 closed Sep 22, 2020

• Installation error

  #43028 closed Sep 22, 2020

• Keras load model throwing no custom loss function while loading so used compile=False, but still the same error

  #42778 closed Sep 22, 2020

• I have found this error. tensorflow.python.framework.errors_impl.InvalidArgumentError: indices[5] = 1 is not in [0, 1)

  #43376 closed Sep 22, 2020

• Random NaN loss when using float16 dtype and batch size of 1

  #43172 closed Sep 22, 2020

• Using Tensorflow-2.3.0 with GPU

  #43337 closed Sep 22, 2020

• how to save weights without optimizer weights by using checkpoint file format in tensorflow keras model?

  #43451 closed Sep 22, 2020

• Neural machine translation with attention issue :Translate function give Keyerror with Tunisian dialect

  #43032 closed Sep 22, 2020

• Can't loading model from checkpoint if use adam optimizer

  #42953 closed Sep 22, 2020

• tf.data.experimental.service throws error when using with TPUs

  #43125 closed Sep 22, 2020

• TFLite Shared Library build error on Raspberry Pi 4 (native build with Bazel). Trying to build with XNNPACK enabled

  #42471 closed Sep 22, 2020

• Huge size difference of GPU delegate library between static and dynamic

  #43102 closed Sep 22, 2020

• How to check the tensor value(this is different from issue 27519)

  #43182 closed Sep 22, 2020

• TFLite: Cannot run inference on TF Lite Model: "Regular TensorFlow ops are not supported by this interpreter."

  #40157 closed Sep 22, 2020

• The command line is too long

  #42601 closed Sep 22, 2020

• keras Sequential model get empty metrics_names

  #43142 closed Sep 22, 2020

• tf.cast with division impose different influence on python2 v.s. python3

  #34686 closed Sep 22, 2020

• Tf Lite Micro Global Average Pooling not supported

  #43332 closed Sep 21, 2020

• GPUDelegate Produces Incorrect Result for reduce_sum

  #43393 closed Sep 21, 2020

• Why is the quantified TF_LITE slower than the unquantified TF_LITE test time on the PC side？

  #43013 closed Sep 21, 2020

• Same variable name for tf.feature_column.embedding_column embedding tensors in tf 2.3.0

  #43011 closed Sep 21, 2020

• TFLite compiling error

  #42892 closed Sep 21, 2020

• TF_lite inference error

  #43010 closed Sep 21, 2020

• LSTM Input Shape Error

  #42986 closed Sep 21, 2020

• Tensorflow 2.0 killed during training (opened graph)

  #42768 closed Sep 21, 2020

• Save checkpoints and continue to training

  #42852 closed Sep 21, 2020

• Some of the operators in the model are not supported by the standard TensorFlow Lite runtime. If those are native TensorFlow operators, you might be able to use the extended runtime by passing --enable_select_tf_ops, or by setting target_ops=TFLITE_BUILTINS,SELECT_TF_OPS when calling tf.lite.TFLiteConverter(). Otherwise, if you have a custom implementation for them you can disable this error with --allow_custom_ops, or by setting allow_custom_ops=True when calling tf.lite.TFLiteConverter(). Here is a list of builtin operators you are using: ADD, CAST, CONCATENATION, CONV_2D, DIV, EXPAND_DIMS, FILL, FLOOR, GATHER, GATHER_ND, LEAKY_RELU, MAXIMUM, MEAN, MINIMUM, MUL, PACK, PAD, RANGE, RESHAPE, RESIZE_BILINEAR, RESIZE_NEAREST_NEIGHBOR, SHAPE, STRIDED_SLICE, SUB, TRANSPOSE_CONV, UNPACK. Here is a list of operators for which you will need custom implementations: MaxPoolWithArgmax, Size.

  #42912 closed Sep 21, 2020

• ValueError: slice index 0 of dimension 0 out of bounds. for 'strided_slice' (op: 'StridedSlice') with input shapes: [0], [1], [1], [1] and with computed input tensors: input[1] = <0>, input[2] = <1>, input[3] = <1>.

  #42091 closed Sep 21, 2020

• tensorflow.python.framework.errors_impl.InvalidArgumentError: Cannot convert a Tensor of dtype resource to a NumPy array.

  #42113 closed Sep 21, 2020

• Tensorflow flooding my Spyder Console

  #42736 closed Sep 21, 2020

• Documentation is becoming too undecipherable for a meaningful practical work

  #42896 closed Sep 21, 2020

• ImportError: DLL load failed: A dynamic link library (DLL) initialization routine failed.

  #42947 closed Sep 21, 2020

• Cannot Download MNIST on WSL2 - Temporary Failure in Name Resolution

  #42985 closed Sep 21, 2020

• tf.GradientTape().gradient() returns None

  #42984 closed Sep 21, 2020

• Issues while installing for raspberry pi through pip command

  #43006 closed Sep 21, 2020

• Error when build the pip package for Tensorflow 1.8

  #43004 closed Sep 21, 2020

• ValueError: No gradients provided for any variable (Keras 2.4, Tensorflow 2.3.0)

  #42038 closed Sep 21, 2020

• Tensorflow 2.3: No gradients provided for any variable

  #43132 closed Sep 21, 2020

• tflite_runtime Interpreter import error

  #43087 closed Sep 21, 2020

• tf.lite.Optimize.DEFAULT - Hybrid models are not supported on TFLite Micro.

  #43386 closed Sep 20, 2020

• How to fix "AttributeError: module 'tensorflow.keras.backend' has no attribute 'get_session'"

  #42981 closed Sep 20, 2020

• Memory leak in model.fit

  #37505 closed Sep 20, 2020

• Usage of intermediate layer causes an exception

  #43293 closed Sep 20, 2020

• RNN compatibility issue on unknown batch_size

  #27941 closed Sep 19, 2020

• fit gets slower on consecutive calls when model is recompiled

  #25073 closed Sep 19, 2020

• [TF 2.0 API Docs] tf.keras.activations.softmax

  #26372 closed Sep 19, 2020

• tensorflow import error

  #43361 closed Sep 19, 2020

• TextVectorization not working on TPU with tf-nightly.

  #43254 closed Sep 19, 2020

• Better Instruction/Insight as to how models.load_model() custom objects are loaded (doc/feature request/bug?)

  #42830 closed Sep 19, 2020

• Different gradients in tf2 when eager mode is enabled compared to graph mode

  #43349 closed Sep 18, 2020

• Colab, TPU training using keras InternalError: Failed to serialize message

  #34321 closed Sep 18, 2020

• Using custom model with fewer that 1000 classes

  #42948 closed Sep 18, 2020

• I got error

  #42926 closed Sep 18, 2020

• Regarding the operation of arbitrary scale access to the network

  #42950 closed Sep 18, 2020

• A problem about the install of tensorflow

  #42810 closed Sep 18, 2020

• SavedModel exporting fails on RNNs by setting wrong input dtypes

  #41815 closed Sep 18, 2020

• g++/gcc cannot find header files in/or libtensorflow_cc.so

  #43307 closed Sep 18, 2020

• “default Round and Robin”, “tf.fixed_ size_ Partitioner” and “tf.min_max_variable_partitioner”

  #42675 closed Sep 18, 2020

• Tensorflow 2.3.0 Java API on Windows - UnsatisfiedLinkError: Cannot find TensorFlow native library for OS

  #42891 closed Sep 18, 2020

• Unexpected behaviour for model.evaluate inside keras Callback

  #43262 closed Sep 18, 2020

• tf.io.gfile.listdir is inconsistent between GCS dir and local dir - adds trailing slashes

  #39167 closed Sep 18, 2020

• A/libc: Fatal signal 11 (SIGSEGV), code 1, fault addr 0x70 in tid 28007

  #34313 closed Sep 18, 2020

• Model and layers set to non-trainable, but weights still adjusted

  #43304 closed Sep 18, 2020

• 'retval_' must have the same nested structure in the main and else branches

  #42575 closed Sep 18, 2020

• "Tensor is unhashable" and "too many values to unpack" with transformers

  #41204 closed Sep 18, 2020

• Bandwidth on TPU v3 > 900 GB/s

  #42566 closed Sep 18, 2020

• [tf1.15] Low GPU usage with multi gpus and multi losses

  #42881 closed Sep 18, 2020

• Redundant functions between `Tensorflow Features Columns` API and `TF Keras Preprocessing Layers`

  #35680 closed Sep 18, 2020

• Int16 softmax is using std::vector

  #43126 closed Sep 17, 2020

• Embedding layer doesn't follow mixed_precision policy by default

  #42631 closed Sep 17, 2020

• [tflite2.0] tf.lite.Interpreter fails with 'post-training quantization' (tf.matmul).

  #33233 closed Sep 17, 2020

• A bug in 'MoveBinaryOperatorBeforeReshape' graph_transformations during tflite_convert.

  #33190 closed Sep 17, 2020

• Can't get MnasNet mean & stdev for input normalization: get correct classes but incorrect scores.

  #31464 closed Sep 17, 2020

• Segfault on absurdly large shuffle buffer for `make_csv_dataset`

  #43119 closed Sep 17, 2020

• ModelCheckpoint results in Input 0 of layer is incompatible with the layer

  #38729 closed Sep 17, 2020

• Keras fails to feed ragged/sparse inputs with correct input placeholder

  #34157 closed Sep 17, 2020

• A dynamic link library (DLL) initialization routine failed.

  #43130 closed Sep 17, 2020

• Usage model's intermediate layer output in custom loss causes OOM

  #41200 closed Sep 17, 2020

• conv2d_transpose param tensor shapes differ from conv2d shapes

  #31192 closed Sep 17, 2020

• reduce_prod EXTREMELY slow

  #23901 closed Sep 17, 2020

• [Regression] on_train_batch_begin callbacks with no batch number and size

  #40703 closed Sep 17, 2020

• FP16 not working with NHNet

  #43273 closed Sep 17, 2020

• Metrics passed to on_batch_end(...) in Keras are inconsistent

  #36400 closed Sep 17, 2020

• GatherV2 checks batch_dims falsely in graph mode

  #42522 closed Sep 17, 2020

• TFlite Converter Error: from tensorflow.lite.toco.python.toco_from_protos import main ModuleNotFoundError: No module named 'tensorflow'

  #40121 closed Sep 16, 2020

• TensorFlow nightly build 2.4.0.dev20200816 or newer breaks Horovod (0.19.1 and newer) installation.

  #42547 closed Sep 16, 2020

• 2020-07-14 nightlies cannot import `tensorflow`

  #41388 closed Sep 16, 2020

• Tensorboard file writer not working any more after nightly build tf-nightly 1.14.1.dev20190410

  #27713 closed Sep 16, 2020

• I want a example code to use the embedding projector in tensorboard by tensorflow-2.0.

  #32879 closed Sep 16, 2020

• How preprocess image from tf.io.decode_raw ?

  #42899 closed Sep 16, 2020

• Does TensorFlow1.x support CUDA11? And how can i compile tensorflow 1.x with cuda11?

  #42895 closed Sep 16, 2020

• Helm chart for TF2+GPU

  #43264 closed Sep 16, 2020

• custom xception input size broadcasting incorrectly

  #43162 closed Sep 16, 2020

• MobileNetV3 Small/Large not fully quantized after full integer post-training quantization runs with no errors

  #43257 closed Sep 16, 2020

• tf.GradientTape().jacobian() triggers retracing

  #42835 closed Sep 16, 2020

• Class weights issue with sparse data from tf.data.dataset where y.shape.rank is None

  #43248 closed Sep 16, 2020

• tf2.3 c++库编译失败

  #43234 closed Sep 16, 2020

• Importing Tensorflow after using pyinstaller (ImportError: DLL load failed while importing _pywrap_tensorflow_internal: The specified module could not be found.)

  #43259 closed Sep 16, 2020

• tf.keras.Model.compile unexpectedly recovered old optimizer iterations

  #42877 closed Sep 16, 2020

• Tensorflow 2.3 not detecting gpu (cuda 10.0)

  #43189 closed Sep 16, 2020

• Data augmentation on TfRecords for multichannel images

  #43145 closed Sep 16, 2020

• Usage of MethodNameUpdater() function

  #43233 closed Sep 16, 2020

• I am not able to install tensorflow

  #43244 closed Sep 16, 2020

• Non responsive model when building micro speech with cmsis-nn

  #42918 closed Sep 16, 2020

• NXP FRDM K66F example for micro speech does not even build

  #43001 closed Sep 16, 2020

• set_floatx('float16') slow to build/compile

  #35553 closed Sep 16, 2020

93 Issues opened by 88 people

• Additive Attention Layer is not really Additive Attention

  #43471 opened Sep 23, 2020

• Bad Loss History, Callback vs. Progressbar Loss Mismatch

  #43470 opened Sep 22, 2020

• reduce_sum and reduce_prod Permuting order of columns on Power9 CPUs

  #43469 opened Sep 22, 2020

• [ROCm] Incorrect path to clang.

  #43468 opened Sep 22, 2020

• "Cadence processor cores only" license restriction

  #43467 opened Sep 22, 2020

• Is there any user data that is either being collected or Tracked by the tensorflow

  #43465 opened Sep 22, 2020

• 'Tensor' object has no attribute 'numpy'

  #43464 opened Sep 22, 2020

• Flower_Classification_with_TFLite_Model_Maker: missing labels.txt

  #43463 opened Sep 22, 2020

• g++ compilation results in undefined reference to `tensorflow::internal::LogMessageFatal::LogMessageFatal - collect2: error: ld returned 1 exit status

  #43461 opened Sep 22, 2020

• Feature Request: Support for ragged argmin/argmax

  #43460 opened Sep 22, 2020

• ModuleNotFoundError: No module named 'tensorflow'

  #43459 opened Sep 22, 2020

• Very simplistic embedding projector example needed

  #43456 opened Sep 22, 2020

• A commit message should be "unit64"

  #43455 opened Sep 22, 2020

• Feature request: Test in cl_build for the esp32

  #43450 opened Sep 22, 2020

• Complete functionality of SparseTensor.__mul__

  #43448 opened Sep 22, 2020

• top_k crashes for certain large structures

  #43447 opened Sep 22, 2020

• TF 2.3 concrete function spec issue: list index out of range

  #43436 opened Sep 21, 2020

• nan is clipped to the upper

  #43433 opened Sep 21, 2020

• import error -> tensorflow

  #43431 opened Sep 21, 2020

• ValueError: Failed to convert value into readable tensor.

  #43430 opened Sep 21, 2020

• Warning when trying to Load CIFAR10 in Tensorflow 1.x

  #43429 opened Sep 21, 2020

• Negate Kernel for int8 is yet to be implemented

  #43425 opened Sep 21, 2020

• Cannot fit concatenated model to two datasets as input

  #43423 opened Sep 21, 2020

• ValueError: Can only save/restore ResourceVariables when executing eagerly

  #43422 opened Sep 21, 2020

• Unable to build Tensorflow 2.3 on OSX with LLVM

  #43421 opened Sep 21, 2020

• I encountered the following problems when installing the Go version of TensorFlow in the windows environment

  #43420 opened Sep 21, 2020

• invalid syntax error while importing tensorflow and keras

  #43419 opened Sep 21, 2020

• Got Segmentation fault when calling tflite::Model::UnPack()

  #43416 opened Sep 21, 2020

• In load_model method: Input 'values' passed int64 expected int32

  #43409 opened Sep 21, 2020

• Hexagon delegate performance

  #43397 opened Sep 20, 2020

• So frustrating with this error: Failed to load the native TensorFlow runtime

  #43387 opened Sep 20, 2020

• TF 2.3: Learning rate is NOT printed in the training output when using "ReduceLROnPlateau" callback

  #43385 opened Sep 20, 2020

• Using keras GRUCell layer in TFLM

  #43380 opened Sep 20, 2020

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

  #43379 opened Sep 20, 2020

• google colab crashed after run tensofrflow code which is upgraded from 1.8 to 2.2

  #43378 opened Sep 20, 2020

• Extract Graph in TensorFlow 2

  #43377 opened Sep 20, 2020

• XLA future

  #43374 opened Sep 20, 2020

• TF2.0, how to include feature processing code in saved model.

  #43373 opened Sep 20, 2020

• Build from source for the Raspberry Pi: Bazel binary wasn't found

  #43372 opened Sep 20, 2020

• whats the equivalent of torch.nn.Parameter() in TF ?

  #43371 opened Sep 20, 2020

• Can lhlo_fuse_linalg support fusion like which in XLA ?

  #43370 opened Sep 20, 2020

• Cant save keras RNN model with custom cell whose call function accepts constants

  #43369 opened Sep 19, 2020

• Tim

  #43368 opened Sep 19, 2020

• Unable to build tensorflowlite.dll in Windows with FLEX delegate support - library limit of 65535 objects exceeded

  #43367 opened Sep 19, 2020

• C++ compilation of rule '@nccl_archive//:device_lib' failed

  #43364 opened Sep 19, 2020

• Module 'tensorflow' has no attribute 'get_default_session'

  #43363 opened Sep 19, 2020

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

  #43362 opened Sep 19, 2020

• TFLITE not enable RUY AVX runtime path

  #43360 opened Sep 19, 2020

• micro_speech.bin file creation fails even if compiling is successful

  #43359 opened Sep 19, 2020

• cuda and tensorflow

  #43352 opened Sep 18, 2020

• tf.contrib.lookup.string_to_index_table_from_file optimization

  #43351 opened Sep 18, 2020

• TF2.0 SavedModel with feature processing inference code

  #43350 opened Sep 18, 2020

• r1.15 Eigen download link is dead

  #43348 opened Sep 18, 2020

• Documentation about tensorflow

  #43347 opened Sep 18, 2020

• Support AWS IAM roles when using TensorFlow file_io

  #43344 opened Sep 18, 2020

• Deprecated get_losses_for function in migration guide

  #43341 opened Sep 18, 2020

• Tensorflow lite speech sample is not detecting silence (wrong detections)

  #43340 opened Sep 18, 2020

• Tensorflow Lite: iOS 14 breaks the NPU Delegate

  #43339 opened Sep 18, 2020

• 2020-09-18 19:08:48.694940: W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'cudart64_101.dll'; dlerror: cudart64_101.dll not found 2020-09-18 19:08:48.699968: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.

  #43336 opened Sep 18, 2020

• Output of TensorFlow Java API differs from the one in TensorFlow Python

  #43335 opened Sep 18, 2020

• F tensorflow/core/framework/tensor_shape.cc:44] Check failed: NDIMS == dims() (2 vs. 5)Asking for tensor of 2 dimensions from a tensor of 5 dimensions

  #43334 opened Sep 18, 2020

• InternalError:cudnn poolforward launch failed

  #43331 opened Sep 18, 2020

• How does TF decide to add examples for one API?

  #43330 opened Sep 18, 2020

• Standard Deviation calculation Pytorch vs TF, how to set unbiased

  #43324 opened Sep 18, 2020

• AttributeError: '_TfDeviceCaptureOp' object has no attribute 'node_def'

  #43323 opened Sep 18, 2020

• vectorized_map fails if function contains an abs of a complex128

  #43322 opened Sep 18, 2020

• Undocumented NotFoundError occurs when using tf.io.gfile.glob() to retrieve files in a non-existent directory

  #43319 opened Sep 17, 2020

• TensorFlow Custom loop gives different value of loss

  #43312 opened Sep 17, 2020

• Building Tensorflow using non-default GCC fails with /lib64/libstdc++.so.6: version `GLIBCXX_3.4.20' not found and further errors

  #43311 opened Sep 17, 2020

• tf.nn.softmax_cross_entropy_with_logits output on multichannel continuous targets

  #43309 opened Sep 17, 2020

• How about we stop deprecating everything every other release....

  #43303 opened Sep 17, 2020

• changes its APIs quite often

  #43302 opened Sep 17, 2020

• Should we remove deprecated and ignored field 'version' in graph.proto

  #43301 opened Sep 17, 2020

• Crash when attempting to use xla.conv with complex inputs

  #43300 opened Sep 17, 2020

• There is a bug about tf/keras/Sequential#fit()

  #43298 opened Sep 17, 2020

• Kernel crash with complex matrices on Windows

  #43297 opened Sep 17, 2020

• Arguments to max_pool are missing correspondance

  #43295 opened Sep 17, 2020

• How to evluate a Model accuracy

  #43294 opened Sep 17, 2020

• Didnt find op for builtin opcode 'RESIZE_NEAREST_NEIGHBOR' version '3'

  #43291 opened Sep 17, 2020

• ERROR: Could not find a version that satisfies the requirement tensorflow==1.14.0

  #43289 opened Sep 17, 2020

• edgeTPU compiler error on TF2.3 quantized model

  #43288 opened Sep 17, 2020

• My doc build is picking up methods not in the advertised API

  #43278 opened Sep 16, 2020

• Failed to run on the given Interpreter: Regular TensorFlow ops are not supported by this interpreter. Make sure you apply/link the Flex delegate before inference. Node number 62011 (FlexSize) failed to prepare.

  #43277 opened Sep 16, 2020

• Model trains on GPU but not on TPU

  #43276 opened Sep 16, 2020

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

  #43275 opened Sep 16, 2020

• Can't Access Tensorflow Dependencies After Install

  #43272 opened Sep 16, 2020

• TFLite converter aborts (dumps core) in BroadcastAdd4DSlow

  #43271 opened Sep 16, 2020

• [RNN]Cannot get saved .tflite model

  #43270 opened Sep 16, 2020

• TF 2.3 - loading of saved_model from disk with ragged=True input is slower than ragged=False

  #43263 opened Sep 16, 2020

• Self-attention on word embeddings using half-precision with mask_zero set to True crashes training

  #43261 opened Sep 16, 2020

• ValueError: No gradients provided for any variable: ['embedding_2/embeddings:0', 'bidirectional_4/forward_lstm_4/lstm_cell_13/kernel:0', 'bidirectional_4/forward_lstm_4/lstm_cell_13/recurrent_kernel:0', 'bidirectional_4/forward_lstm_4/lstm_cell_13/bias:0', 'bidirectional_4/backward_lstm_4/lstm_cell_14/kernel:0', 'bidirectional_4/backward_lstm_4/lstm_cell_14/recurrent_kernel:0', 'bidirectional_4/backward_lstm_4/lstm_cell_14/bias:0', 'bidirectional_5/forward_lstm_5/lstm_cell_16/kernel:0', 'bidirectional_5/forward_lstm_5/lstm_cell_16/recurrent_kernel:0', 'bidirectional_5/forward_lstm_5/lstm_cell_16/bias:0', 'bidirectional_5/backward_lstm_5/lstm_cell_17/kernel:0', 'bidirectional_5/backward_lstm_5/lstm_cell_17/recurrent_kernel:0', 'bidirectional_5/backward_lstm_5/lstm_cell_17/bias:0', 'dense_2/kernel:0', 'dense_2/bias:0'].

  #43258 opened Sep 16, 2020

• mAP drops crazily when replacing Conv with Depthwise_Conv

  #43256 opened Sep 16, 2020

• Failed to load the native TensorFlow runtime.

  #43255 opened Sep 16, 2020

271 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.

• Adding per-replica dataset distribution

  #38968 commented on Sep 22, 2020 • 26 new comments

• Clarified the DispatchServer creation process in tf.data distribute service docs

  #43023 commented on Sep 22, 2020 • 17 new comments

• Docker with GPU 2.3rc0 CUDA runtime implicit initialization on GPU:0 failed. Status: device kernel image is invalid

  #41132 commented on Sep 18, 2020 • 11 new comments

• Didn't find op for builtin opcode 'CONV_2D' version '5'

  #43232 commented on Sep 21, 2020 • 9 new comments

• Adding HDFS connection cache on tensorflow side can improve performance by 20+ times #43187

  #43188 commented on Sep 22, 2020 • 9 new comments

• [Intel MKL] Adding support for Conv3D and Depthwise Conv with native format

  #43213 commented on Sep 17, 2020 • 9 new comments

• java.lang.IllegalArgumentException: Internal error: Failed to apply delegate: Attempting to use a delegate that only supports static-sized tensors with a graph that has dynamic-sized tensors

  #38036 commented on Sep 18, 2020 • 6 new comments

• Tensorflow 2.2 and 2.3 not detecting GPU with CUDA 10.1

  #43236 commented on Sep 22, 2020 • 6 new comments

• W tensorflow/stream_executor/platform/default/dso_loader.cc:59] Could not load dynamic library 'cudart64_110.dll'; dlerror: cudart64_110.dll not found

  #43193 commented on Sep 22, 2020 • 6 new comments

• Add __reduce_ex__ to Keras Model to enable copy.deepcopy and pickle

  #39609 commented on Sep 21, 2020 • 6 new comments

• DataFormatDimMap supports 5D tensors and grappler layout optimizer supports ReduceOps for 5D tensors

  #43226 commented on Sep 22, 2020 • 6 new comments

• QAT conversion RuntimeError: Quantization not yet supported for op: 'DEQUANTIZE' issue with tf-nightly

  #42082 commented on Sep 20, 2020 • 5 new comments

• Failed to convert SparseTensor to Tensor

  #42916 commented on Sep 22, 2020 • 5 new comments

• Model conversion fails with cryptic error when feeding sample during post-training quantization

  #43239 commented on Sep 22, 2020 • 5 new comments

• [Intel MKL] Enabling BatchNorm test in AutoMPMkl

  #41476 commented on Sep 22, 2020 • 5 new comments

• [TFLu] Update initialization of 'input_dims' in EvalQuantizedInt8() method of Fully Connected kernel (CMSIS-NN)

  #42314 commented on Sep 22, 2020 • 5 new comments

• [ROCm] Fix for ROCm CSB Breakage - 200902 - 1

  #42897 commented on Sep 21, 2020 • 5 new comments

• MUL doesn't work on gpu delegate with error: Dimension can not be reduced to linear

  #37102 commented on Sep 18, 2020 • 4 new comments

• Training of a simple model with KerasLayer hangs on macOSX with tf 2.3.0

  #43060 commented on Sep 19, 2020 • 4 new comments

• android use XNNPACK happen error

  #42056 commented on Sep 20, 2020 • 4 new comments

• Unable to access files on S3 with tf.io.gfile.GFile

  #38054 commented on Sep 20, 2020 • 4 new comments

• tensorflow/core/grappler/optimizers/meta_optimizer.cc:561] layout failed: Invalid argument: MutableGraphView::SortTopologically error:

  #36901 commented on Sep 21, 2020 • 4 new comments

• RuntimeError: Quantization not yet supported for op: %

  #43207 commented on Sep 22, 2020 • 4 new comments

• Keras on_epoch_end() callback output happens BEFORE completion of epoch output with verbose=1

  #43184 commented on Sep 22, 2020 • 4 new comments

• TFLite model gives random outputs when running on Android

  #43071 commented on Sep 23, 2020 • 4 new comments

• CallCounter C++ version

  #42540 commented on Sep 19, 2020 • 4 new comments

• [T.F 2.0 API Docs] Modifying python math_ops.py documentation

  #42619 commented on Sep 22, 2020 • 4 new comments

• CMSIS-NN conv, depthwise conv kernels: allocate per channel memory dynamically (#42748)

  #42770 commented on Sep 22, 2020 • 4 new comments

• Autograph applied to Keras Custom Loss during Eager Execution

  #41189 commented on Sep 16, 2020 • 3 new comments

• tflite delagte to nn accelerator

  #42826 commented on Sep 17, 2020 • 3 new comments

• Could not find TensorFlow Java 2.3.0 release in maven central

  #42670 commented on Sep 18, 2020 • 3 new comments

• The inconsistency of CTC API. Please optimize it.

  #42993 commented on Sep 21, 2020 • 3 new comments

• Understanding the Difference between Normal Training vs eager mode training in TFOD API v2

  #43097 commented on Sep 22, 2020 • 3 new comments

• Cannot confine TensorFlow C API to use not more than 1 threads in total

  #42510 commented on Sep 22, 2020 • 3 new comments

• libtensorflow_inference.so: protobuf failed to link __android_log_write and dl_iterate_phdr

  #43235 commented on Sep 22, 2020 • 3 new comments

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

  #43202 commented on Sep 22, 2020 • 3 new comments

• Support "fetch_skip_sync=false" in Callable

  #33187 commented on Sep 22, 2020 • 3 new comments

• Loading a quantized TFlite model fails when allocating tensors

  #38311 commented on Sep 23, 2020 • 3 new comments

• Make target for generic cortex M4F device

  #41860 commented on Sep 18, 2020 • 3 new comments

• Update nn_ops.py

  #42324 commented on Sep 22, 2020 • 3 new comments

• Update tensorflow.keras.metrics.pbtxt

  #42442 commented on Sep 22, 2020 • 3 new comments

• [TFLite 16x8] Fixes for TANH and LOGISTIC

  #42671 commented on Sep 22, 2020 • 3 new comments

• Request a new padding mode

  #18213 commented on Sep 17, 2020 • 2 new comments

• get_output_shapes does not return the correct dimensions for a Dataset

  #43250 commented on Sep 17, 2020 • 2 new comments

• tf.keras.callbacks.TensorBoard histogram summary breaks on tf.bool layer weights

  #43200 commented on Sep 18, 2020 • 2 new comments

• dense_to_ragged_batch fails with map function implemented with py_function

  #43096 commented on Sep 19, 2020 • 2 new comments

• Pybind11 exception with tensorflow 2.2 env in python cpp communication

  #42405 commented on Sep 19, 2020 • 2 new comments

• TensorFlow hangs forever in multinode training with NCCL and certain model (with SyncBatchNorm layers)

  #42111 commented on Sep 21, 2020 • 2 new comments

• NotFoundError: No algorithm worked! when using Conv2D

  #43174 commented on Sep 21, 2020 • 2 new comments

• Maybe incorrect Brier score calculation

  #41664 commented on Sep 22, 2020 • 2 new comments

• Unable to build tflite aar file on windows

  #42258 commented on Sep 22, 2020 • 2 new comments

• error when trying to modify code to train on tpu's.

  #43186 commented on Sep 22, 2020 • 2 new comments

• Models loaded from files do not perform as before saving

  #43230 commented on Sep 22, 2020 • 2 new comments

• imx8 tflite opencl

  #43063 commented on Sep 22, 2020 • 2 new comments

• interperter with GPU Delegate blocks Android from rendering on Main thread.

  #43073 commented on Sep 22, 2020 • 2 new comments

• Tensorflow build from source doesn’t see GPUs gets “ failed call to cuInit: UNKNOWN ERROR (-1)”

  #41748 commented on Sep 22, 2020 • 2 new comments

• Golang Tensorflow v2.3.0 installation fails

  #41808 commented on Sep 23, 2020 • 2 new comments

• [TFLite] Fix potential overflow in 64-bit MultiplyByQuantizedMultiplier function

  #42134 commented on Sep 23, 2020 • 2 new comments

• Add SegmentProdGrad

  #42237 commented on Sep 19, 2020 • 2 new comments

• [40171] Add VarHandleOp to set of kernels not to cache

  #42377 commented on Sep 20, 2020 • 2 new comments

• [ROCm] Removing TPU tests from the ROCm CSB + CI scripts.

  #42656 commented on Sep 22, 2020 • 2 new comments

• [XLA] Upstream master grid size

  #42683 commented on Sep 21, 2020 • 2 new comments

• [ROCm] Switch rocm build to use ROCm 3.7

  #42689 commented on Sep 22, 2020 • 2 new comments

• [ROCm] Fix for ROCm CSB Breakage - 200902 - 2

  #42944 commented on Sep 21, 2020 • 2 new comments

• Enable FusedBatchNormV3 ops to support 5D tensors

  #42970 commented on Sep 22, 2020 • 2 new comments

• [TF-TRT] Adding AMP support for non converted OPs & native segments

  #42974 commented on Sep 23, 2020 • 2 new comments

• [T.F 2.0 API Docs] Modifying documentation of set_optimizer in config.py

  #43105 commented on Sep 19, 2020 • 2 new comments

• tf.sparse.reduce_sum slower/less memory efficient than (unsorted_)segment_sum

  #32763 commented on Sep 16, 2020 • 1 new comment

• How to replace a layer of a compiled Keras' model

  #36320 commented on Sep 16, 2020 • 1 new comment

• Deep Training with Tensorflow

  #37044 commented on Sep 16, 2020 • 1 new comment

• DLL load failed: A dynamic link library (DLL) initialization routine failed.

  #43169 commented on Sep 16, 2020 • 1 new comment

• Interpreter AllocateTensors() does not finish or report error

  #43238 commented on Sep 16, 2020 • 1 new comment

• fatal error: tensorflow/core/protobuf/error_codes.pb.h: No such file or directory

  #42721 commented on Sep 16, 2020 • 1 new comment

• Integer quantization error

  #43129 commented on Sep 16, 2020 • 1 new comment

• setting verbosity to 0 does not suppress keras debug messages, no way to turn them off

  #43253 commented on Sep 16, 2020 • 1 new comment

• TF_lite Convert using integer-only quantization

  #43062 commented on Sep 16, 2020 • 1 new comment

• TFX for Mobile: Additional dtypes

  #43045 commented on Sep 16, 2020 • 1 new comment

• No gradients provided for any variable

  #43065 commented on Sep 16, 2020 • 1 new comment

• TensorBoard callback without profile_batch setting cause Errors: CUPTI_ERROR_INSUFFICIENT_PRIVILEGES and CUPTI_ERROR_INVALID_PARAMETER

  #35860 commented on Sep 16, 2020 • 1 new comment

• tf.argmin across longer axis significantly slower than tf.reduce_min

  #42521 commented on Sep 16, 2020 • 1 new comment

• TensorFlow Lite Op Request

  #34150 commented on Sep 16, 2020 • 1 new comment

• Dataset iterator hanging from within a dataset interleave

  #34659 commented on Sep 16, 2020 • 1 new comment

• tf.nn.fractional_max_pool aborts when pooling_ratio is negative

  #42278 commented on Sep 16, 2020 • 1 new comment

• ValueError: Unable to save the object ListWrapper(...) (a list wrapper constructed to track trackable TensorFlow objects) when calling the method tf.keras.Model.save_weights

  #36916 commented on Sep 16, 2020 • 1 new comment

• Unknown: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above.

  #36025 commented on Sep 16, 2020 • 1 new comment

• Does the framework support IPv6 networks?

  #42921 commented on Sep 16, 2020 • 1 new comment

• ModelCheckpoint callback can't save entire subclassed models

  #37839 commented on Sep 16, 2020 • 1 new comment

• Error when computing gradients of a reloaded SavedModel containing an if clause

  #40795 commented on Sep 16, 2020 • 1 new comment

• Keras model.predict on large datasets throws OOM

  #42862 commented on Sep 16, 2020 • 1 new comment

• ValueError: Unknown floatx type: bfloat16 tf.keras.backend.set_floatx('bfloat16')

  #37926 commented on Sep 16, 2020 • 1 new comment

• Request for Prebuilt C++ Library libtensorflow_cc.so (probably from CI artifacts)

  #42614 commented on Sep 17, 2020 • 1 new comment

• How to implement data-dependent initialization with MultiGPU/TPU [Tensorflow >= 2.x]

  #39645 commented on Sep 17, 2020 • 1 new comment

• Pre-trained weights channel ordering clarification.

  #42430 commented on Sep 17, 2020 • 1 new comment

• model.predict is much slower on TF 2.1+

  #40261 commented on Sep 17, 2020 • 1 new comment

• tf.keras.callbacks.ProgbarLogger(count_mode='samples') does not work

  #38765 commented on Sep 17, 2020 • 1 new comment

• [tflite] Processing time is high in Windows 10 but better in Ubuntu 16.0

  #43061 commented on Sep 17, 2020 • 1 new comment

• tf-mlir-translate is not coming in binaries

  #43203 commented on Sep 17, 2020 • 1 new comment

• tensorflow lite div quantized problem

  #42882 commented on Sep 17, 2020 • 1 new comment

• OpenCL

  #43171 commented on Sep 17, 2020 • 1 new comment

• Generating ICU normalization_data.c and icu_conversion_data.c on s390x architecture

  #43124 commented on Sep 17, 2020 • 1 new comment

• I want to add a "RecvTensor" Op on "timeline" Tracing for RDMA , But The code does not worked, can help me ?

  #43116 commented on Sep 17, 2020 • 1 new comment

• Tensorflow profiler crashes when using string categorical layer

  #41244 commented on Sep 17, 2020 • 1 new comment

• DepthwiseConv1D: implementation, check and warning message

  #36935 commented on Sep 17, 2020 • 1 new comment

• benchmark error cannot run in phone

  #42956 commented on Sep 17, 2020 • 1 new comment

• Basic Queries regarding MobileBERT predictions using TFLite Model Maker

  #43112 commented on Sep 17, 2020 • 1 new comment

• OP_REQUIRES failed at reshape_op.h:57 : Invalid argument: Size 1 must be non-negative, not -9

  #43092 commented on Sep 17, 2020 • 1 new comment

• Inference only works if using same batch size as in training

  #42864 commented on Sep 17, 2020 • 1 new comment

• TFLite GPU execution failed

  #35632 commented on Sep 17, 2020 • 1 new comment

• Not able to load a tf.keras model

  #38346 commented on Sep 18, 2020 • 1 new comment

• tf.keras.Model#load_weights() cannot be used on a directory generated by tf.keras.callback.ModelCheckpoint

  #38745 commented on Sep 18, 2020 • 1 new comment

• Confused which version TensorfFlow CPU supports

  #43229 commented on Sep 18, 2020 • 1 new comment

• TFLite not support Dynamic input size

  #24607 commented on Sep 18, 2020 • 1 new comment

• tensorflow 1.14 gpu is not in 2060S 2080TI

  #43138 commented on Sep 18, 2020 • 1 new comment

• Problem saving and loading Keras Model that was built with StellarGraph library

  #43114 commented on Sep 18, 2020 • 1 new comment

• voice activity detection implementation

  #28757 commented on Sep 18, 2020 • 1 new comment

• Unrecognized command line -msse3 while cross-compiling tensorflow to aarch64 targets

  #42686 commented on Sep 18, 2020 • 1 new comment

• Model Input & Output does not match.

  #43242 commented on Sep 18, 2020 • 1 new comment

• Tensorflow V2.2.0 boot fail on aarch64

  #42545 commented on Sep 18, 2020 • 1 new comment

• TF 2.3 Saved model | Experimental features not working properly

  #42613 commented on Sep 18, 2020 • 1 new comment

• Tried to export a function which references untracked object Tensor("5486:0", shape=(), dtype=resource).TensorFlow objects (e.g. tf.Variable) captured by functions must be tracked by assigning them to an attribute of a tracked object or assigned to an attribute of the main object directly.

  #40310 commented on Sep 18, 2020 • 1 new comment

• tf.signal.inverse_stft segfault when frame_length is a large value

  #42246 commented on Sep 18, 2020 • 1 new comment

• tensorflow.keras slower than keras for repetitive calls to Model.predict()

  #40412 commented on Sep 18, 2020 • 1 new comment

• Bug using dictionary in tensorflow2.3

  #42679 commented on Sep 18, 2020 • 1 new comment

• Tensorflow Java API in Scala with scala-reflect runtime compilation

  #37787 commented on Sep 19, 2020 • 1 new comment

• KeyError: 'Failed to format this callback filepath: "checkpoint_5000/checkpoint_{epoch:02d}_{batch:04d}". Reason: \'batch\''

  #40907 commented on Sep 19, 2020 • 1 new comment

• ValueError: Graph disconnected: cannot obtain value for tensor Tensor…The following previous layers were accessed without issue:

  #42017 commented on Sep 19, 2020 • 1 new comment

• tf.keras.applications.EfficientNetB- not running as expected

  #42506 commented on Sep 19, 2020 • 1 new comment

• when i am running the sample tutorial code of basic text classification i just cannot download a file named imdb_word_index.json

  #43245 commented on Sep 19, 2020 • 1 new comment

• Count number of times `tf.function` is traced

  #37323 commented on Sep 19, 2020 • 1 new comment

• Unable to cross compile TFLite for Raspberry Pi Zero using Docker image nightly-develop

  #38037 commented on Sep 19, 2020 • 1 new comment

• Error computing gradient of a vectorized scan loop

  #41789 commented on Sep 20, 2020 • 1 new comment

• Slot variables used in an optimizer must have the same shape with the variable to be optimized?

  #19457 commented on Sep 20, 2020 • 1 new comment

• Allow slots to be initialized with different shapes + values than primary var in Keras optimizer

  #41675 commented on Sep 20, 2020 • 1 new comment

• Cuda 10: Failed to get device properties, error code: 3

  #26572 commented on Sep 20, 2020 • 1 new comment

• Tensorflow crashes on source building!!!

  #23436 commented on Sep 20, 2020 • 1 new comment

• Unexpected crash when loading a modified saved_model.pb file

  #41097 commented on Sep 20, 2020 • 1 new comment

• AttributeError: 'BatchDataset' object has no attribute 'ndim'

  #42855 commented on Sep 21, 2020 • 1 new comment

• Cannot convert a Tensor of dtype resource to a NumPy array

  #42673 commented on Sep 21, 2020 • 1 new comment

• Discrepancy between loss when called from `tf.keras.losses.BinaryCrossentropy()(true, pred)` and verbose output from model.fit() or hand-computation

  #43143 commented on Sep 21, 2020 • 1 new comment

• Building and testing CPU Ops,

  #43022 commented on Sep 21, 2020 • 1 new comment

• Issue on Load Trax

  #43086 commented on Sep 21, 2020 • 1 new comment

• LSTM Issue

  #43090 commented on Sep 21, 2020 • 1 new comment

• tensorflow:AutoGraph could not transform <function Model.make_train_function.<locals>.train_function at 0x0000018DDF556708> and will run it as-is.

  #43154 commented on Sep 21, 2020 • 1 new comment

• TFLite speech example failing to train with -output_representation='spec'or'mfcc' examples/lite/examples/speech_commands/ml

  #30682 commented on Sep 21, 2020 • 1 new comment

• TF_TensorToPyArray is hard coded for equality, limits performance enhancements

  #43044 commented on Sep 21, 2020 • 1 new comment

• How can I clear GPU memory in tensorflow 2?

  #36465 commented on Sep 21, 2020 • 1 new comment

• Segfault with tf.linalg.cholesky

  #43170 commented on Sep 21, 2020 • 1 new comment

• ValueError: name for name_scope must be a string when Building up a Custom Model

  #35337 commented on Sep 21, 2020 • 1 new comment

• InternalError : failed to call ThenRNNBackward... on version 2.3 but not 2.1

  #42412 commented on Sep 21, 2020 • 1 new comment

• "Failed starting model allocation." while trying to allocate tensors

  #43084 commented on Sep 21, 2020 • 1 new comment

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

  #42972 commented on Sep 21, 2020 • 1 new comment

• ValueError: Tensor-typed variable initializers must either be wrapped in an init_scope or callable (e.g., `tf.Variable(lambda : tf.truncated_normal([10, 40]))`) when building functions.

  #41306 commented on Sep 21, 2020 • 1 new comment

• Successful NUMA node read from SysFS had negative value (-1)

  #42738 commented on Sep 21, 2020 • 1 new comment

• Full integer Quantization of SSD-Mobilenet V2 doesn't work

  #42426 commented on Sep 21, 2020 • 1 new comment

• Min_delta parameters on callbacks function of Keras model seems to not take effect

  #41417 commented on Sep 21, 2020 • 1 new comment

• From Keras 2.3.1 to 2.4: Input 0 of layer fc1 is incompatible with the layer

  #41537 commented on Sep 21, 2020 • 1 new comment

• Can you optimize the text.text_to_word_sequence API？

  #41647 commented on Sep 21, 2020 • 1 new comment

• Please update the tflite_convert -h command line help

  #43027 commented on Sep 22, 2020 • 1 new comment

• Undefined symbols for architecture arm64 when loading TensorFlowLiteSelectTfOps on iOS device

  #41948 commented on Sep 22, 2020 • 1 new comment

• Please release TFLite wheels for Beaglebone

  #42906 commented on Sep 22, 2020 • 1 new comment

• LR Custom Scheduler not working TF 2.3

  #43219 commented on Sep 22, 2020 • 1 new comment

• micro speech example not working with STM32F769

  #43201 commented on Sep 22, 2020 • 1 new comment

• [TFLu] int8 ops slower than f32

  #43121 commented on Sep 22, 2020 • 1 new comment

• "Function call stack: _dist_train_step" error on tensorflow2 object detection api

  #43228 commented on Sep 22, 2020 • 1 new comment

• "Function call stack: _dist_train_step" error on tensorflow2 object detection api

  #43227 commented on Sep 22, 2020 • 1 new comment

• tensorflow model to MLIR

  #43212 commented on Sep 22, 2020 • 1 new comment

• Issue with Training CenterNet Resnet50 V2 Keypoints 512x512

  #43179 commented on Sep 22, 2020 • 1 new comment

• Error converting multilingual universal sentence encoder to TFLite. Input 1 of node StatefulPartitionedCall was passed float from statefulpartitionedcall_args_1:0 incompatible with expected resource.

  #42366 commented on Sep 22, 2020 • 1 new comment

• The New Converter of Tensorflow 2.2.0 generate incorrect model output

  #39572 commented on Sep 22, 2020 • 1 new comment

• [r2.0.0-rc1] Converting to TFLite format: Invalid quantization params for op RESHAPE

  #32495 commented on Sep 22, 2020 • 1 new comment

• tf.data.experimental.make_csv_dataset, currently only supports one label; However, most datasets can have more than one label

  #43083 commented on Sep 22, 2020 • 1 new comment

• tf.compat.v1.data.get_output_shapes(datasets) gives Errors

  #42824 commented on Sep 22, 2020 • 1 new comment

• scaling issue by using multiworkerstrategy for CPU

  #42888 commented on Sep 22, 2020 • 1 new comment

• Memory leak with tf.py_function in eager mode using TF 2.0

  #35010 commented on Sep 22, 2020 • 1 new comment

• [TFlite]: Failed to instantiate the interpreter with a StyleGAN2 generator model

  #42190 commented on Sep 22, 2020 • 1 new comment

• tf.string_split() cannot speedup with num_parallel_calls increased in Dataset.map()

  #42917 commented on Sep 22, 2020 • 1 new comment

• AttributeError: 'dict' object has no attribute 'name'

  #38988 commented on Sep 22, 2020 • 1 new comment

• ConnectionResetByPeer using MultiWorkerMirroredStrategy with Native Keras BERT Model

  #42498 commented on Sep 23, 2020 • 1 new comment

• Feature Request: Pass 'elapsed_secs' into LoggingHook string formatter.

  #34308 commented on Sep 23, 2020 • 1 new comment

• [New Feature] tf.data pipeline serialization to json for performance analysis

  #37517 commented on Sep 17, 2020 • 1 new comment

• [tf.image.convert_image_dtype] - Add extra warning if scaling is skipped

  #39233 commented on Sep 23, 2020 • 1 new comment

• Fix invalid fusion of Matmul and Mul

  #41790 commented on Sep 22, 2020 • 1 new comment

• Adding methods and tests for ndarray: ndarray.sum(), ndarray.argmax(), ndarray.argmin(), ndarray.any(), ndarray.all(), ndarray.clip(), ndarray.diagonal().

  #42289 commented on Sep 18, 2020 • 1 new comment

• Fix deprecated usage of collections ABC

  #42338 commented on Sep 22, 2020 • 1 new comment

• Add EuclideanNorm XLA kernel

  #42385 commented on Sep 18, 2020 • 1 new comment

• Add CMSIS-NN SVDF kernel.

  #42651 commented on Sep 18, 2020 • 1 new comment

• [TFLite 16x8, documentation] Added section on 16x8 quantization scheme to model-optimization.

  #42699 commented on Sep 22, 2020 • 1 new comment

• [WIP] Disable TFTRT conversion of ResizeBilinear Nodes when align_corners=false

  #42901 commented on Sep 22, 2020 • 1 new comment

• [Intel MKL] unit test for bfloat16 on matmul.

  #42949 commented on Sep 16, 2020 • 1 new comment

• Rename libtensorflowlite.a to libtensorflow-lite.a

  #43144 commented on Sep 22, 2020 • 1 new comment

• [TFLite] Ensure that all int16x8 operators check that the zero-point is null

  #43148 commented on Sep 22, 2020 • 1 new comment

• Tweaks to build tensorflow-lite-select-tf-ops

  #43198 commented on Sep 18, 2020 • 1 new comment

• Integrate cuBLASLt API into backend

  #43237 commented on Sep 19, 2020 • 1 new comment

• TFLu: Fix Ethos-U build errors

  #43240 commented on Sep 22, 2020 • 1 new comment

• [tflite] Conv2DTranspose output difference between mobile CPU and mobile GPU

  #42649 commented on Sep 16, 2020 • 0 new comments

• Feature request: MLIR-based TFLite converter support for 16bit conv2d

  #43247 commented on Sep 16, 2020 • 0 new comments

• Question about TFLite GPU delegate

  #43208 commented on Sep 16, 2020 • 0 new comments

• tf.keras.applications.ResNet50 does not work properly

  #43194 commented on Sep 16, 2020 • 0 new comments

• Tensorflow hub USE model issue

  #43190 commented on Sep 16, 2020 • 0 new comments

• Cannot import name 'image_preprocessing'

  #43147 commented on Sep 16, 2020 • 0 new comments

• tensorflow.python.framework.errors_impl.UnknownError:

  #43133 commented on Sep 16, 2020 • 0 new comments

• TF 2.3 ImportError: DLL load failed: The specified module could not be found.

  #43085 commented on Sep 16, 2020 • 0 new comments

• tf.debugging.set_log_device_placement(True) does not seem to be working correctly.

  #42708 commented on Sep 16, 2020 • 0 new comments

• Improve tf.data graph representation in TensorBoard

  #21349 commented on Sep 16, 2020 • 0 new comments

• tf.keras.callbacks.TensorBoard does not save batch - level statistics when profiling is disabled completely

  #42139 commented on Sep 16, 2020 • 0 new comments

• keras.callbacks.Tensorboard cannot reduce loss and produces error

  #36646 commented on Sep 16, 2020 • 0 new comments

• AttributeError: module 'tensorboard' has no attribute 'lazy'

  #37433 commented on Sep 16, 2020 • 0 new comments

• TensorFlowOpLayer messes up the TensorBoard graphs / grouping with TensorBoard

  #39398 commented on Sep 16, 2020 • 0 new comments

• tf.keras/Tensorboard callback does not implement activation histogram (and says it does)

  #39755 commented on Sep 16, 2020 • 0 new comments

• Keras Tensorboard Callback conflicts with Lambda Callback writing summaries

  #39730 commented on Sep 16, 2020 • 0 new comments

• TPUStrategy does not export graph to TensorBoard while TPUEstimator does it

  #40372 commented on Sep 16, 2020 • 0 new comments

• Stale(?) references to activation histograms

  #42027 commented on Sep 16, 2020 • 0 new comments

• Graph visualization failed, in Graph mode [TF 2.2, TB 2.2.1]

  #42133 commented on Sep 16, 2020 • 0 new comments

• Cannot use dictionary embeddings metadata for keras callbacks

  #33967 commented on Sep 16, 2020 • 0 new comments

• Tracing of first batch does not collect compute events

  #33571 commented on Sep 16, 2020 • 0 new comments

• Embedding visualization in tf.keras / TensorFlow 2.0

  #33230 commented on Sep 16, 2020 • 0 new comments

• un-deprecate `write_grads` for `fit`

  #31173 commented on Sep 16, 2020 • 0 new comments

• tf.contrib.training.HParams domain interval support

  #26436 commented on Sep 16, 2020 • 0 new comments

• Tensorboard embeddings mnist example crashes

  #30115 commented on Sep 16, 2020 • 0 new comments

• Slow iteration/converting of tensors to vector<float> via Python C API

  #42120 commented on Sep 16, 2020 • 0 new comments

• Embedding for Embedding Projector works on one computer, not on another.

  #42380 commented on Sep 16, 2020 • 0 new comments

• Undocumented S3_DISABLE_MULTI_PART_DOWNLOAD variable and its behavior

  #42389 commented on Sep 16, 2020 • 0 new comments

• Sparse tensor error message when applying constraint to dense variable

  #43205 commented on Sep 17, 2020 • 0 new comments

• OP_REQUIRES failed : Not found: No algorithm worked!

  #43223 commented on Sep 17, 2020 • 0 new comments

• SyncBatchNormalization has NaN losses with channels-first format

  #43165 commented on Sep 17, 2020 • 0 new comments

• tf.keras SparseCategoricalCrossentropy with sample_weight on TPU: Error DenseToDenseSetOperation (No registered 'DenseToDenseSetOperation)

  #43195 commented on Sep 17, 2020 • 0 new comments

• How to use unified memory in TF2.1

  #43049 commented on Sep 17, 2020 • 0 new comments

• tf.nn.max_pool2d: No support for tf.Tensor input for ksize

  #42973 commented on Sep 17, 2020 • 0 new comments

• No gradient defined for operation RaggedTensorFromVariant / or no gradients at all

  #43107 commented on Sep 17, 2020 • 0 new comments

• nan gradient issue

  #42889 commented on Sep 18, 2020 • 0 new comments

• tf.data.Dataset API with ImageGenerator => ValueError: as_list() is not defined on an unknown TensorShape.

  #41843 commented on Sep 18, 2020 • 0 new comments

• Allow symmetric TFLite quantization (no zero point/scale only)

  #43099 commented on Sep 18, 2020 • 0 new comments

• Saver not created because there are no variables in the graph to restore

  #40824 commented on Sep 18, 2020 • 0 new comments

• Convert saved model - issue generated shapes

  #42153 commented on Sep 19, 2020 • 0 new comments

• Adding HDFS connection cache on tensorflow side can improve performance by 20+ times

  #43187 commented on Sep 19, 2020 • 0 new comments

• FusedBatchNorm of TF 1.9 doesn't work fine

  #21328 commented on Sep 21, 2020 • 0 new comments

• Support CUDA 10.2

  #39247 commented on Sep 21, 2020 • 0 new comments

• Quantization aware training does not add weights quantization and fake quantization nodes to Conv2D layers without bias and followed by non-addition ops

  #31153 commented on Sep 22, 2020 • 0 new comments

• Checkpoint is not work properly

  #39651 commented on Sep 22, 2020 • 0 new comments

• tensorflow lite native app, undefined reference to `NnApiImplementation()'

  #42276 commented on Sep 22, 2020 • 0 new comments

• Generate sparse tensor in tflite

  #43035 commented on Sep 22, 2020 • 0 new comments

• A quantized model for TensorFlow Lite is slow.

  #43128 commented on Sep 22, 2020 • 0 new comments

• TensorFlow Lite causes segementation faults in C++ when creating threads if fully statically linked

  #42057 commented on Sep 22, 2020 • 0 new comments

• Using intermediate layer outputs in custom loss function causes CUDA_ERROR_OUT_OF_MEMORY

  #33837 commented on Sep 22, 2020 • 0 new comments

• High CPU memory usage when calling GradientTape's gradient() when using multiple threads/cores

  #41718 commented on Sep 22, 2020 • 0 new comments

• Loss follows Learning Rate trend when using tf.keras.optimizers.schedules.LearningRateSchedule

  #41794 commented on Sep 22, 2020 • 0 new comments

• Embedding layer ~10x slower in tf.function

  #38287 commented on Sep 22, 2020 • 0 new comments

• Add Usage Example to Keras Nadam Optimizer

  #35984 commented on Sep 16, 2020 • 0 new comments

• Added ResNext models

  #37146 commented on Sep 16, 2020 • 0 new comments

• [ROCm] Fixing and enabling TopK

  #39429 commented on Sep 17, 2020 • 0 new comments

• Add missing double overloads for GetAttr, GetNodeAttr, and TryGetNodeAttr

  #39452 commented on Sep 23, 2020 • 0 new comments

• [TF-TRT] Fix TopK Combined NMS

  #40062 commented on Sep 18, 2020 • 0 new comments

• [TFLite] Added op tests for conv_activations

  #41344 commented on Sep 22, 2020 • 0 new comments

• Patch to core for RFC: Sparse Domain Isolation for Supporting Large-scale Sparse Weights

  #41371 commented on Sep 19, 2020 • 0 new comments

• [TFLite] Add int16x8 support for BATCH_MATMUL operator

  #42059 commented on Sep 21, 2020 • 0 new comments

• Windows symbol exports for C++ API

  #42063 commented on Sep 16, 2020 • 0 new comments

• Make ./configure work with Ubuntu's nvidia-cuda-toolkit package

  #42187 commented on Sep 16, 2020 • 0 new comments

• [TF:MLIR] Add TF to TF lowering patterns to legalize_tf pass for TFLite

  #42525 commented on Sep 22, 2020 • 0 new comments

• Update to callbacks.py EarlyStopping

  #42813 commented on Sep 16, 2020 • 0 new comments

• [INTEL MKL] Remove nGraph build support

  #42870 commented on Sep 21, 2020 • 0 new comments

• Add ImageProjectiveTransform XLA kernel

  #42871 commented on Sep 16, 2020 • 0 new comments

• Update CMSIS-NN archive to support SVDF.

  #43059 commented on Sep 22, 2020 • 0 new comments

• TFLu: Don't construct PoolParams if not needed in CMSIS int8 kernel

  #43067 commented on Sep 16, 2020 • 0 new comments

• Pywrap ops

  #43088 commented on Sep 16, 2020 • 0 new comments

• [Intel MKL] Fixing mkl-eager-op-rewrite-test

  #43091 commented on Sep 16, 2020 • 0 new comments

• [TFLite] Add int16x8 support for ABS operator

  #43098 commented on Sep 22, 2020 • 0 new comments

• [ROCm] Disabling failing unit tests

  #43159 commented on Sep 17, 2020 • 0 new comments

• Uses of the new rocm skip test decorator.

  #43161 commented on Sep 16, 2020 • 0 new comments

• Support 5D tensors in the binary ops of Grappler Layout Optimizer

  #43166 commented on Sep 22, 2020 • 0 new comments

• Make fast builds work with MSVC

  #43178 commented on Sep 16, 2020 • 0 new comments

• [Intel MKL] Adding support for convolution fusions with native format

  #43216 commented on Sep 23, 2020 • 0 new comments