Weekly GitHub Report for Tensorflow: November 03, 2025 - November 10, 2025 (12:08:45)

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
  • 1.2. Other Noteworthy Updates
• II. Issues
  • 2.1. Top 5 Active Issues
  • 2.2. Top 5 Stale Issues
  • 2.3. Open Issues
  • 2.4. Closed Issues
  • 2.5. Issue Discussion Insights

• III. Pull Requests
  • 3.1. Open Pull Requests
  • 3.2. Closed Pull Requests
  • 3.3. Pull Request Discussion Insights

• IV. Contributors
  • 4.1. Contributors

I. News

1.1 Recent Version Releases:

The current version of this repository is v2.19.0

1.2 Version Information:

Released on March 5, 2025, TensorFlow version 2.19.0 introduces breaking changes to the tf.lite API, including the deprecation of tf.lite.Interpreter in favor of ai_edge_litert.interpreter and changes to certain C++ constants for improved API compatibility. Key updates also include runtime support for the bfloat16 data type in the tfl.Cast operation, alongside the discontinuation of standalone libtensorflow package publishing.

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. Bot comments are omitted.

1. Can't download tensorflow: This issue concerns a user unable to download TensorFlow on macOS 15 due to using Python 3.14, which is not yet supported by the current TensorFlow release 2.20. The user encounters an error indicating no matching TensorFlow distribution is found for their Python version, and seeks advice on resolving the compatibility problem.

   • The comments clarify that TensorFlow 2.20 supports up to Python 3.13, recommending the user downgrade their Python version to 3.13 or lower to successfully install TensorFlow. Additional unrelated advice about upgrading typing_extensions and virtual environment activation for a different setup was also shared, but the main resolution centers on Python version compatibility.
   • Number of comments this week: 2

2. Fix the doc of axis and Raises in func tf.linspace: This issue concerns a documentation bug in the TensorFlow function tf.linspace, specifically regarding the inaccurate description of the axis argument and the missing Raises section in the docstring. The reporter highlights that the current documentation incorrectly states that axis is only used for N-D tensors, which does not align with the actual behavior that causes an error when used improperly, and suggests updating the argument description and adding the exceptions raised.

   • The comments confirm that a pull request addressing the documentation fixes has been submitted and is under review, with maintainers expressing gratitude for the contribution and indicating that merging the PR will resolve the issue.
   • Number of comments this week: 2

3. build tflite error for android arm64: This issue reports a build error encountered when compiling TensorFlow Lite for the Android arm64 platform using a specific commit. The error messages indicate multiple undefined types and identifiers related to the XNNPACK delegate's file_util.cc source file, suggesting missing or mismatched declarations in the code.

   • The single comment suggests that the XNNPACK component might be missing some header files, implying that adding the necessary headers could resolve the build errors.
   • Number of comments this week: 1

4. How does TensorFlow manage the lifecycle of an input tensor that is used within a GPU Kernel?: This issue questions how TensorFlow manages the lifecycle of an input tensor used within a GPU kernel, specifically inquiring about the absence of explicit CUDA stream synchronization in certain methods and the potential risk of accessing deallocated memory if the CUDA kernel has not completed execution. The user seeks clarification on whether there is an internal mechanism that prevents such race conditions or memory access errors during tensor deallocation after an operation finishes.

   • The single comment reiterates the concern about the lack of explicit CUDA stream synchronization in TensorFlow v2.5 methods, highlighting the risk of race conditions or memory access violations if the CUDA kernel execution is incomplete when the input tensor is deallocated.
   • Number of comments this week: 1

5. Inconsistent error message when using tf.function with NumPy arrays in eager mode: This issue reports a bug where using tf.function with NumPy arrays in eager execution mode results in an unclear error message that only states “Graph execution error” without specifying that the input should be a Tensor rather than a NumPy array. The user provides a minimal reproducible example and notes that the error occurs in TensorFlow version 2.16.0 on Ubuntu with Python 3.10.12.

   • A commenter tested the provided code in TensorFlow 2.16 on Colab and did not encounter the reported error, sharing a notebook link for reference and asking for clarification if anything was missed.
   • Number of comments this week: 1

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.

1. TF-TRT Warning: Could not find TensorRT: This issue describes a problem where TensorFlow on Ubuntu 22.04 cannot detect TensorRT despite having a compatible NVIDIA RTX 3050 Ti GPU and CUDA 12.4 installed, with the user having to downgrade their NVIDIA driver from version 550 to 535 to maintain system stability. The user reports persistent errors and warnings related to TensorRT not being found, despite multiple attempts to reinstall TensorFlow and CUDA components, and is seeking assistance to resolve this configuration and compatibility issue.
2. SystemError in tf.ensure_shape and tf.compat.v1.ensure_shape when dtype of shape is tf.uint64 and its value is too large.: This issue reports a bug in TensorFlow where using tf.ensure_shape or tf.compat.v1.ensure_shape with a shape tensor of type tf.uint64 containing very large values close to 2^64 causes a SystemError and OverflowError. Specifically, when such large values are passed in eager execution mode, the functions fail with an error related to the built-in isinstance function, indicating improper handling of large unsigned 64-bit integers in shape validation.
3. Feature Request: Integrate different Digital Signal Processing into tf.signal: This issue is a feature request proposing the integration of advanced digital signal processing (DSP) functionalities, similar to those found in the julius library, into TensorFlow's tf.signal module. The requester highlights the current lack of sophisticated audio data augmentation tools within TensorFlow compared to PyTorch and suggests that adding these capabilities would enhance audio model training by enabling native, efficient preprocessing and augmentation workflows.
4. [DOCS] Missing complex input for Round op: This issue reports a documentation bug in TensorFlow where the Round operation is described as supporting complex tensor inputs, but in practice, attempting to use a complex tensor with this operation results in an error, requiring users to manually round the real and imaginary parts separately. The user provides a reproducible example on MacOS with TensorFlow 2.15.0 and Python 3.9, demonstrating that the operation fails with a device-not-found error despite the documentation indicating support for complex types.
5. tf.raw_ops.Unbatch aborts with "Check failed: d < dims()": This issue reports a bug in TensorFlow version 2.17 where the tf.raw_ops.Unbatch operation aborts with a "Check failed: d < dims()" error, causing the program to crash. The problem occurs when using certain random tensor inputs on a Linux Ubuntu 20.04.3 system with Python 3.11.8, and it has been reproduced with TensorFlow Nightly builds.

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: 67

Summarized Issues:

• CUDA kernel memory errors in TensorFlow 2.20: Multiple issues report illegal out-of-bounds memory reads and writes in various CUDA kernels such as Bucketize, UnaryClipCustomKernel, DebugNumericSummaryV2, DenseBincount, DynamicStitch, MaxPoolGradGradWithArgmax, MaxPoolBackward, and MirrorPadGrad. These errors are detected by Compute-Sanitizer on Ubuntu 22.04 with CUDA 12.5.1 and cuDNN 9.2.1, causing crashes and undefined behavior during GPU execution.
• [issues/103469, issues/103470, issues/103994, issues/103995, issues/103996, issues/103997, issues/103998, issues/103999]

• MaxPoolWithArgmax operation bugs: There are two related bugs where tf.nn.max_pool_with_argmax works on CPU but fails on GPU due to missing OpKernel registration for XLA_GPU_JIT and an error caused by unsupported output_dtype in backpropagation gradients. These issues cause graph compilation failures and contradict the documented behavior of the function.
• [issues/103438, issues/103454]

• TensorFlow Lite build and linking issues: Building TensorFlow Lite for Android arm64 fails due to undefined identifiers and mismatched declarations in xnnpack delegate source files. Additionally, on Linux, linking applications using system-installed gRPC and TensorFlow Lite's libtensorflow-lite.so causes conflicts due to different absl-cpp versions, and the TensorFlow Lite installation lacks a package config file, preventing proper build script discovery.
• [issues/103429, issues/103536, issues/103538]

• Multi-GPU training performance overheads: Using TensorFlow's MirroredStrategy for multi-GPU training results in significant performance inefficiencies including poor memory copy overlap, GPU compute start time disparities, increased compute time per GPU, and pinned memory limits affecting host-to-device transfers. These overheads degrade training speed compared to single GPU usage.
• [issues/103564]

• TensorFlow CUDA compatibility and installation issues: TensorFlow 2.20.0 running on Kali Linux under WSL2 with NVIDIA RTX PRO Blackwell GPUs and CUDA 12.8 suffers from compatibility problems due to missing CUDA kernel binaries for the GPU's compute capability, causing JIT compilation delays. The official macOS installation guide is missing instructions to install tensorflow-macos and tensorflow-metal packages, leading to segmentation faults during verification.
• [issues/103531, issues/104016]

• Requests for improved CUDA/cuDNN dependency management: There is a feature request to automate detection and installation of compatible CUDA and cuDNN versions for TensorFlow and PyTorch GPU support, addressing the current manual configuration process that often leads to version mismatches and GPU initialization failures.
• [issues/103532]

• Documentation corrections and clarifications: Several issues request updates to TensorFlow documentation, including correcting error condition descriptions for tf.linalg.LinearOperatorScaledIdentity, tf.linalg.LinearOperatorCirculant, and tf.linalg.LinearOperatorToeplitz, clarifying tf.raw_ops.RFFT2D behavior regarding fft_length, and improving tf.linspace parameter documentation and error raising sections.
• [issues/103640, issues/103660, issues/103911, issues/103915, issues/103920]

• TensorFlow import and runtime errors: Users report an ImportError on Windows with TensorFlow 2.19.0 and Python 3.12.9 due to DLL load failure in the native runtime, and a bug in TensorFlow 2.20.0 where combining RAII mutex_lock with manual unlock/lock calls causes double-unlock errors leading to undefined behavior during checkpoint loading.
• [issues/103918, issues/103924]

• TensorFlow Lite usage and build guidance requests: Users request samples or documentation for the recently added external buffer support feature in TensorFlow Lite and seek guidance on using CMake and compiled TensorFlow Lite files to create a demo after building the project.
• [issues/103559, issues/103630]

• TensorFlow symbol visibility and build errors: TensorFlow 2.20.0 has a broken symbol visibility control causing libtensorflow_framework.so to expose excessive dynamic symbols including internal LLVM symbols, leading to symbol conflicts with other LLVM-based libraries and crashing the Python process. Additionally, building TensorFlow 2.20.0 on macOS with Xcode 26.1 fails due to missing LC_UUID load command errors.
• [issues/104038, issues/103590]

• TensorFlow function input handling and error messaging: TensorFlow 2.16.0 produces unclear error messages when using tf.function in eager mode with NumPy array inputs, only stating "Graph execution error" without specifying the cause, which confuses users.
• [issues/103823]

• Comprehensive airline customer service contact information: Numerous issues provide detailed official phone numbers, multilingual support, 24-hour availability, and guidance for contacting airlines in Brazil including LATAM, United, KLM, Turkish Airlines, Qatar Airways, Air Europa, Emirates, American Airlines, Lufthansa, Air Canada, Copa Airlines, Iberia, Air France, Delta, British Airways, and JetBlue. These cover services such as reservations, cancellations, refunds, baggage, flight changes, special assistance, loyalty programs, and emergencies.
• [issues/103945, issues/103946, issues/103947, issues/103948, issues/103950, issues/103961, issues/103962, issues/103964, issues/103965, issues/103966, issues/103967, issues/103968, issues/103969, issues/103970, issues/103971, issues/103972, issues/103973, issues/103974, issues/103975, issues/103976, issues/103977, issues/104025, issues/104027, issues/104028, issues/104030, issues/104032, issues/104035, issues/104036, issues/104037, issues/104039, issues/104040, issues/104041, issues/104042, issues/104043]

2.4 Closed Issues

This section lists, groups, and then summarizes issues that were closed within the last week in the repository. This section also links the associated pull requests if applicable.

Issues Closed This Week: 14

Summarized Issues:

• TensorFlow Model Training Crash: This issue describes a sporadic and environment-specific fatal exception causing a core dump during the .fit() method of a TensorFlow model training session on TensorFlow 2.20.0 with Python 3.13.7 on Ubuntu 25.10. The maintainers could not reproduce the problem, indicating it may be related to the user's specific setup.
• issues/103430

• Spam Issues Posting Airline Customer Service Contacts: Multiple issues repeatedly post detailed contact information and instructions for reaching various airline customer service centers in Brazil, including KLM, Latam, Air France, Turkish Airlines, United Airlines, JetBlue, Air Europa, and Emirates. These issues were closed by maintainers as spam because they are unrelated to the TensorFlow project and contain unsolicited customer service details.
• issues/103617, issues/103618, issues/103835, issues/103838, issues/103839, issues/103840, issues/103841, issues/103842, issues/103848, issues/103849, issues/103850, issues/103851, issues/103872

2.5 Issue Discussion Insights

This section will analyze the tone and sentiment of discussions within this project's open and closed issues that occurred within the past week. It aims to identify potentially heated exchanges and to maintain a constructive project environment.

1. ~(“KLM BRASIL TELEFONE”)~ Como posso falar com a KLM Brasil?
   • Toxicity Score: 0.75 (Rapid escalation, spam content, administrative closure)
   • This GitHub conversation involves a single user posting repetitive and irrelevant content that appears to be spam, followed by a maintainer or community member closing the issue with a brief comment labeling it as spam. The tone from the poster is neutral but off-topic, while the responder's tone is firm and administrative. The trigger for tension is the inappropriate nature of the original post, leading to swift moderation action.

---

III. Pull Requests

3.1 Open Pull Requests

This section provides a summary of pull requests that were opened in the repository over the past week. The top three pull requests with the highest number of commits are highlighted as 'key' pull requests. Other pull requests are grouped based on similar characteristics for easier analysis. Up to 25 pull requests are displayed in this section, while any remaining pull requests beyond this limit are omitted for brevity.

Pull Requests Opened This Week: 8

Key Open Pull Requests

1. Fix simplememoryarena null check: This pull request addresses a potential crash in the SimpleMemoryArena::Commit function by adding a null pointer check after memory reallocation failure, ensuring that the function returns an error and logs it appropriately instead of proceeding with a null pointer, thereby preventing undefined behavior during tensor allocation.

• URL: pull/103428

• Merged: No

• Associated Commits: 52856

2. Add better detection for ANE: This pull request proposes adding improved detection for Apple Neural Engine (ANE) by incorporating support for macOS and enhancing detection methods beyond relying solely on the device model number.

• URL: pull/103481

• Merged: No

• Associated Commits: 0396a

3. Fix the doc of axis and Raises in func tf.linspace: This pull request aims to correct the documentation of the axis parameter and the Raises section in the tf.linspace function to address issue #103660.

• URL: pull/103661

• Merged: No

• Associated Commits: 5c011

Other Open Pull Requests

• Documentation corrections for LinearOperator error handling: Multiple pull requests focus on improving the accuracy of documentation related to errors raised by various tf.linalg.LinearOperator classes. These changes ensure that the docstrings for LinearOperatorScaledIdentity, LinearOperatorCirculant, and LinearOperatorToeplitz correctly describe the errors raised, enhancing clarity for users.
• pull/103912, pull/103916, pull/103921

• Go bindings GPUOptions support: One pull request adds support for GPUOptions in the Go bindings by creating a Go representation of GPUOptions and integrating it into SessionOptions.Config. This update allows serialization to ConfigProto and passes the configuration to TF_SetConfig() before session creation, addressing issue tensorflow/tensorflow#22926.
• pull/104004

• Thread safety improvements in TensorSliceReaderCache: A pull request refactors the TensorSliceReaderCache::GetReader function to use scoped RAII locks for automatic mutex management, preventing a potential double-unlock and race condition. This change enhances thread safety without modifying the existing caching and notification logic.
• pull/104023

3.2 Closed Pull Requests

This section provides a summary of pull requests that were closed in the repository over the past week. The top three pull requests with the highest number of commits are highlighted as 'key' pull requests. Other pull requests are grouped based on similar characteristics for easier analysis. Up to 25 pull requests are displayed in this section, while any remaining pull requests beyond this limit are omitted for brevity.

Pull Requests Closed This Week: 4

Key Closed Pull Requests

1. Feature/doctor office assistant booking tzm: This pull request introduces features for a doctor office assistant, including a patient booking and follow-up system as well as real-time voice functionality, before migrating the code from the TensorFlow repository to a dedicated doctors-assistant repository.

• URL: pull/103473

• Merged: No

• Associated Commits: 8df48, 7cc10, 5ec66

2. Spam: Minor spammy update by Vishnupriya: This pull request is a minor documentation update submitted by Vishnupriya that was identified as spam and was not merged.

• URL: pull/103411

• Merged: No

• Associated Commits: 7571f, 569e3

3. Add dtype note for IsotonicRegression op: This pull request adds a detailed documentation note to the tf.raw_ops.IsotonicRegression operation specifying the supported input-output dtype combinations, clarifying that floating-point inputs retain their dtype, 8- and 16-bit integer inputs produce 32-bit float outputs, 32- and 64-bit integer inputs produce 64-bit float outputs, and that unsupported dtype pairs result in a "Could not find device for node" error, thereby improving dtype behavior clarity and addressing issue #103352.

• URL: pull/103408

• Merged: Yes

• Associated Commits: e68e2

Other Closed Pull Requests

• CUDA Stream Synchronization Awareness: This pull request adds a note and warning about the absence of explicit CUDA stream synchronization in the ExecutorState::ProcessSync function to improve awareness of potential synchronization issues. The update aims to inform developers about synchronization behavior to prevent potential bugs related to CUDA stream handling.
• pull/103443

3.3 Pull Request Discussion Insights

This section will analyze the tone and sentiment of discussions within this project's open and closed pull requests that occurred within the past week. It aims to identify potentially heated exchanges and to maintain a constructive project environment.

Based on our analysis, there are no instances of toxic discussions in the project's open or closed pull requests from the past week.

---

IV. Contributors

4.1 Contributors

Active Contributors:

We consider an active contributor in this project to be any contributor who has made at least 1 commit, opened at least 1 issue, created at least 1 pull request, or made more than 2 comments in the last month.

If there are more than 10 active contributors, the list is truncated to the top 10 based on contribution metrics for better clarity.

Contributor	Commits	Pull Requests	Issues	Comments
ashvinashivin0-sketch	0	0	38	0
CodersAcademy006	11	10	0	8
ILCSFNO	6	6	11	6
Venkat6871	0	0	0	23
khteh	0	0	6	3
kshiteej-mali	7	1	0	0
infiWang	6	1	0	1
kokol16	0	0	8	0
shank87414	0	0	8	0
souhil25	4	2	0	1