Weekly GitHub Report for Tensorflow: December 10, 2024 - December 17, 2024
Weekly GitHub Report for Tensorflow
Thank you for subscribing to our weekly newsletter! Each week, we deliver a comprehensive summary of your GitHub project's latest activity right to your inbox, including an overview of your project's issues, pull requests, contributors, and commit activity.
Table of Contents
I. News
1.1 Recent Version Releases:
The current version of this repository is v2.18.0
1.2 Other Noteworthy Updates:
II. Issues
2.1 Top 5 Active Issues:
We consider active issues to be issues that that have been commented on most frequently within the last week.
- Tensorflow Import Not working: This issue involves a user experiencing difficulties importing TensorFlow on a Windows 11 system, despite following various troubleshooting steps such as reinstalling Python and TensorFlow, and trying different Python versions. The user consistently encounters a DLL load failure error, indicating a problem with the dynamic link library initialization when attempting to import TensorFlow.
- The comment section reveals a series of troubleshooting attempts, including suggestions to use different TensorFlow versions, create new virtual environments, and adjust installation paths. Despite these efforts, the user continues to face the same import error, leading to discussions about potential machine-specific issues, processor compatibility, and the possibility of building TensorFlow from source. The conversation also includes sharing environment details and attempts to replicate the issue on other machines, with ongoing support from the community to resolve the problem.
- Number of comments this week: 9
-
Division by zero error at random places if GPU is used: This issue involves a division by zero error occurring randomly when a program using TensorFlow's C API is executed on a GPU, specifically a Quadro RTX 6000, while it runs without issues on a CPU. The problem is challenging to diagnose as it appears to be deeply rooted in TensorFlow or CUDA, with inconsistent behavior even when the program is run multiple times under the same conditions.
- The comments discuss potential collaboration to resolve the issue, with suggestions for a screen-sharing session to demonstrate the problem. There is a recommendation to update incompatible software versions, and attempts to do so have led to further complications with CUDA diagnostics. Despite various tests and adjustments, the error persists, and contributors are encouraged to follow documentation for contributing to the project.
- Number of comments this week: 8
-
Linking an Android static library with TFLite GPU using CMake causes undefined symbol errors and can not get the correct install: This issue involves a bug where linking an Android static library with TensorFlow Lite's GPU delegate using CMake results in undefined symbol errors, preventing the correct installation of the library. The problem occurs on Ubuntu 20.04.6 LTS with TensorFlow version 2.18, and the user has attempted custom code modifications to resolve the issue but continues to face errors related to missing targets and undefined symbols during the build process.
- The comments discuss various aspects of the issue, including questions about setting up the TensorFlow Lite environment and linking necessary libraries. A suggestion is made to link OpenGL ES and EGL libraries to resolve undefined symbol errors, which helps fix the initial problem. However, a new issue arises with the PReLU operation not being supported, leading to further discussion on how to address this in TensorFlow Lite version 2.18.0.
- Number of comments this week: 6
-
error: defining a type within 'offsetof' is a Clang extension [-Werror,-Wgnu-offsetof-extensions]: This issue involves a user encountering a build error while attempting to compile the Selective Framework for iOS using TensorFlow Lite, specifically due to a Clang extension warning related to defining a type within 'offsetof'. The user is seeking assistance to resolve this error, which is preventing the successful completion of the build process.
- The comments discuss troubleshooting steps, including requests for more information about the user's environment and commands used. Suggestions are made to add a specific compiler flag to bypass the warning causing the error. The user is also advised to modify the build script if necessary. The conversation includes apologies for delayed responses and clarifications on the user's goal to reduce the framework size for iOS.
- Number of comments this week: 5
-
tensorflow error in pycharm in ubuntu 22.04: This issue is about a user experiencing an error when running TensorFlow code in PyCharm on Ubuntu 22.04, despite the code being correct and running successfully on other platforms. The user is seeking help to understand why the error occurs and why certain TensorFlow functions are not recognized in their development environment.
- The comments reveal that the user is experiencing an error in PyCharm on Ubuntu, while the code runs fine on other platforms. Another user tested the code on Colab and Ubuntu without issues, suggesting the problem might be specific to the user's setup. The original poster clarifies that the error involves unrecognized functions, despite the code executing correctly, and they seek further assistance to resolve this discrepancy.
- Number of comments this week: 4
2.2 Top 5 Stale Issues:
We consider stale issues to be issues that has had no activity within the last 30 days. The team should work together to get these issues resolved and closed as soon as possible.
As of our latest update, there are no stale issues for the project this week.
2.3 Open Issues
This section lists, groups, and then summarizes issues that were created within the last week in the repository.
Issues Opened This Week: 18
Summarized Issues:
- Division by Zero Error on GPU with TensorFlow's C API: This issue involves a division by zero error occurring unpredictably when using a GPU with TensorFlow's C API. The error does not occur when running the program on a CPU, indicating a potential GPU-specific problem. The issue persists across different hardware setups and CUDA versions, making it difficult to pinpoint whether the bug lies within TensorFlow, CUDA libraries, or elsewhere.
- Annoying Warning Message in TensorFlow 2.17+: This issue describes an annoying warning message, "Ignoring Assert operator," that appears multiple times during the
model.fit()operation in TensorFlow 2.17+ with Keras 3.x. The warning disrupts the console output by interrupting the progress bar. Attempts to suppress it using environment variables have been unsuccessful.
- Documentation Discrepancy in
tf.raw_ops.MaxPoolGradWithArgmax: This issue highlights a documentation bug in TensorFlow'stf.raw_ops.MaxPoolGradWithArgmaxfunction. The official documentation states that theargmaxargument can be of data typeint32orint64. However, actual testing reveals that onlyint64is supported, leading to a discrepancy between the documentation and the function's behavior.
- Parameter Swap Bug in ImageDataGenerator: This issue highlights a bug in TensorFlow 2.18.0 where the 'height_shift_range' and 'width_shift_range' parameters in the ImageDataGenerator are swapped. This causes 'height_shift_range' to move images left and right and 'width_shift_range' to move images up and down. This behavior is contrary to their intended functionality.
- Compatibility Problem with TensorFlow Lite on Raspberry Pi Zero: This issue involves a compatibility problem with the cross-compilation toolchain for building TensorFlow Lite on a Raspberry Pi Zero. The provided toolchain targets ARMv7 architecture, which is incompatible with the ARMv6 architecture of the Raspberry Pi Zero. This results in an "Illegal instruction" error when attempting to run the compiled binary.
- Publication Request for TensorFlowLite 2.18.0: This issue requests the publication of TensorFlowLite version 2.18.0 to Maven Central and CocoaPods Specs for Android and iOS platforms. The request is made before transitioning to LiteRT. This publication is crucial for developers relying on these platforms for their applications.
- Compilation Error with TensorFlow Lite and ndk-build: This issue involves a compilation error encountered when using ndk-build for a custom C++ file. The error message indicates an undefined symbol related to the
tflite::StatefulNnApiDelegate::StatefulNnApiDelegate(tflite::StatefulNnApiDelegate::Options)in TensorFlow Lite. Attempts to set various NNAPI-related parameters have not resolved the issue.
- ImportError in PyCharm with TensorFlow: This issue involves a user experiencing an ImportError due to a DLL load failure when attempting to run TensorFlow on PyCharm. The error may be caused by missing dependencies, incompatible CPU instructions, or incorrect library installations. The user requests further details on the TensorFlow version, environment, and installation steps.
- Outdated NVIDIA Docker Link in TensorFlow Documentation: This issue highlights that the link for installing NVIDIA Docker support on the TensorFlow Docker installation page is outdated. It points to a deprecated project, suggesting that it should be updated to direct users to the NVIDIA Container Toolkit installation guide instead. This update is necessary to ensure users have the correct installation instructions.
- Compatibility Issue with TensorFlow 2.10.0 and Python 2.7.5: This issue highlights a compatibility problem where TensorFlow version 2.10.0 cannot be installed on Python 2.7.5. TensorFlow 2.x requires Python 3.7 or above, leading to difficulties in running the project with the specified versions. Users need to upgrade their Python version to resolve this issue.
- Bug in
tensorflow.experimental.numpy.kronwith Multi-dimensional Arrays: This issue reports a bug in TensorFlow version 2.18.0 where thetensorflow.experimental.numpy.kronfunction fails to work with multi-dimensional arrays. Specifically, when the input tensors have more than two dimensions, it results in an "UnimplementedError" due to unsupported broadcasting. The function works correctly in NumPy, highlighting a discrepancy.
- LSTM Layer Behavior Difference on GPU with XLA: This issue describes a bug where the
tf.keras.layers.LSTMlayer behaves differently on a GPU when using XLA (Accelerated Linear Algebra). The layer results in a failure, whereas it works correctly without XLA and on a CPU regardless of XLA usage. This inconsistency poses challenges for developers relying on GPU acceleration.
- Installation Error for TensorFlow 2.10.0rc4 on Windows 10: This issue is about a user encountering an error when attempting to install TensorFlow version 2.10.0rc4 on Windows 10 using Python 3.13. The installation process fails because no matching distribution is found for the specified version. This suggests a potential compatibility issue with the Python version.
- GRU Function Output Length Discrepancy on GPU: This issue describes a bug in the TensorFlow library where the
keras.layers.GRUfunction produces different output lengths when executed on a GPU compared to a CPU. The GPU output length is unexpectedly larger due to batch unpacking. This discrepancy can lead to unexpected results in model predictions.
- LSTM-based Time Series Forecasting Guide: This issue involves a detailed step-by-step guide for implementing LSTM-based time series forecasting using Python and TensorFlow. The guide covers data preprocessing, model building, training, and evaluation. It serves as a comprehensive resource for developers looking to apply LSTM models to time series data.
- Crash in
LearnedUnigramCandidateSamplerOperation: This issue involves a bug in TensorFlow where theLearnedUnigramCandidateSampleroperation crashes with an "Aborted (core dumped)" error. The crash occurs due to an invalid argument, specifically when thenum_trueattribute is set to an excessively large value, causing an integer overflow. This bug affects the stability of models using this operation.
- Custom Loss Function Bug in TensorFlow 2.18.0: This issue involves a bug where a custom loss function in TensorFlow 2.18.0, when explicitly declared in the model's compile configuration, only receives the first output (bounding box) from a model with multiple outputs. The function is expected to receive both outputs (bounding box and label classification), but it does not, leading to incorrect loss calculations.
- ImportError with TensorFlow 2.10 on Windows 11: This issue involves a user experiencing an ImportError while attempting to install TensorFlow 2.10 on Windows 11 with Python 3.8.18. The error message indicates a conflict with the python38.dll module, suggesting a potential version mismatch or installation problem during the setup process. The user is advised to follow the TensorFlow installation guide closely to resolve the issue.