Weekly GitHub Report for Tensorflow: September 08, 2025 - September 15, 2025 (12:09:01)

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 better API compatibility. Key updates also include runtime support for the bfloat16 data type in the tfl.Cast operation and the discontinuation of standalone libtensorflow package publishing, while still allowing unpacking from PyPI.

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. [GPU][BiasAdd] Kernel launch count overflow with large Dense/bias_add (negative work_element_count → abort): This issue reports a bug in TensorFlow 2.20.0 where the GPU kernel launch count overflows during a BiasAdd operation on very large tensors, causing a fatal check failure due to a negative work_element_count. The problem occurs with both Keras Dense layers using bias and raw tf.nn.bias_add calls when using bfloat16 data type on tensors with billions of elements, leading to crashes in eager execution and potentially under XLA compilation or SavedModel workflows.

   • A user confirmed reproducing the bug on Colab with TensorFlow 2.20.0 and GPU enabled, providing a reference notebook to demonstrate the issue.
   • Number of comments this week: 1

2. TensorFlow/Keras model accumulates system and GPU RAM during training: This issue describes a problem where training a TensorFlow/Keras model causes both system RAM and GPU RAM usage to continuously increase, eventually exhausting memory and terminating the training process. The user is employing TensorFlow 2.19.0 with a custom data generator and a custom embedding loss function, and is unsure why the memory accumulation occurs during training.

   • The single comment suggests trying alternative memory allocators such as jemalloc or TCmalloc by preloading them before running the training script, aiming to address potential memory management issues causing the RAM buildup.
   • Number of comments this week: 1

3. Note and Clarify the Error Raised in tf.raw_ops.ExtractGlimpse() and tf.raw_ops.ExtractGlimpseV2(): This issue addresses a documentation bug in TensorFlow's tf.raw_ops.ExtractGlimpse() and tf.raw_ops.ExtractGlimpseV2() functions, specifically clarifying the error message related to the simultaneous specification of uniform_noise and noise parameters. The current error description is misleading because it does not accurately reflect the condition under which the error is raised, and the issue requests that this clarification be added to the docstrings of both functions.

   • A pull request has been opened to fix the documentation and clarify the error message as discussed in the issue.
   • Number of comments this week: 1

4. Clarify the Error Raised in tf.train.CheckpointManager(): This issue addresses a documentation bug in the TensorFlow API where the error raised by tf.train.CheckpointManager() when checkpoint_interval is set without providing a step_counter is not clearly documented. The reporter demonstrates that the current docstring lacks information about this required parameter, which leads to a ValueError during usage, and suggests that the documentation should explicitly mention this condition.

   • A pull request has been opened to address the documentation gap highlighted in the issue, aiming to clarify the error conditions related to the step_counter parameter in tf.train.CheckpointManager().
   • Number of comments this week: 1

5. Clarify the range of action of select_cols in tf.io.decode_csv() and tf.raw_ops.DecodeCSV(): This issue concerns a documentation bug in TensorFlow's tf.io.decode_csv() and tf.raw_ops.DecodeCSV() functions, specifically regarding the unclear behavior and range of the select_cols parameter. The reporter highlights that select_cols appears to only apply to the records argument and not to record_defaults, which is not documented, causing confusion and a ValueError when their lengths do not match.

   • A pull request has been opened to address the documentation clarification, indicating that the issue is being actively worked on and the confusion around select_cols usage is being resolved.
   • 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, resulting in a warning message "TF-TRT Warning: Could not find TensorRT" despite having compatible NVIDIA drivers and CUDA versions installed. The user reports difficulties with driver versions, specifically needing to downgrade from the default 550 driver to 535 for their RTX 3050 Ti GPU, and has tried reinstalling TensorFlow multiple times without success, seeking assistance to resolve the TensorRT detection failure.
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 a shape like [18446743219011059112, 1] is passed in eager execution mode, the APIs fail with a SystemError related to the built-in isinstance function, indicating improper handling of large unsigned 64-bit integers.
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 PyTorch julius library, into TensorFlow's tf.signal module. The requester highlights the benefits of having native audio data augmentation tools within TensorFlow to improve model robustness and streamline workflows by reducing reliance on external libraries.
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 round the real and imaginary parts separately. The user provides a reproducible example and error logs showing that the operation fails due to a missing device kernel for complex inputs, highlighting a discrepancy between the official documentation and the actual behavior in TensorFlow version 2.15.0.
5. tf.raw_ops.Unbatch aborts with "Check failed: d < dims()": This issue reports a bug in TensorFlow version 2.17 where the operation tf.raw_ops.Unbatch aborts with a fatal check failure error "Check failed: d < dims()". The user has reproduced this problem on Linux Ubuntu 20.04.3 using Python 3.11.8 with custom code, and the error occurs due to an invalid dimension check during the unbatching process, causing the program to crash.

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: 16

Summarized Issues:

• Null Pointer Dereference Bugs in TensorFlow Components: Multiple issues report potential null pointer dereference bugs in TensorFlow's XLA and MLIR components due to missing return statements or unchecked null variables, which can lead to crashes or undefined behavior during execution. These bugs highlight the need for better null checks and code safety in critical TensorFlow internals.
• issues/99907, issues/99909

• Crashes and Fatal Errors in TensorFlow 2.20.0: Several issues describe fatal errors including GPU kernel launch count overflow causing aborts, segmentation faults after PyVista's screenshot method, and memory exhaustion during model training, indicating stability problems in TensorFlow 2.20.0 and 2.19.0 under specific conditions. These problems affect both GPU operations and memory management, leading to crashes or termination of processes.
• issues/99894, issues/99925, issues/100085

• Symbol Export and Linking Conflicts in TensorFlow Shared Libraries: The shared library libtensorflow_framework.so.2 exports unnecessary symbols from dependencies like LLVM, causing symbol resolution conflicts and potential crashes. The proposed solution involves building with -Bsymbolic and using version scripts to restrict symbol exports and improve stability.
• issues/99991

• Documentation Clarifications and Error Message Improvements: Multiple issues request clearer documentation and error messages for TensorFlow functions, including ExtractGlimpse, CheckpointManager, Dequantize, decode_csv, and QuantizedMaxPool. These clarifications aim to improve user understanding of parameter behaviors, error causes, and expected inputs to prevent confusion and misuse.
• issues/100134, issues/100138, issues/100175, issues/100223, issues/100234

• Unexpected Behavior and Bugs in TensorFlow API Functions: Some TensorFlow API functions exhibit unexpected or incorrect behavior, such as tf.nn.depthwise_conv2d working with unsupported dilation and stride combinations, and tf.random.stateless_uniform throwing exceptions with certain algorithm parameters. These bugs contradict official documentation or expected functionality.
• issues/100237, issues/100252

• User Support and Diagnostic Requests for TensorFlow Lite on Android: An issue requests detailed system information, SDK versions, and logs to diagnose a problem with TensorFlow Lite in Google Play Services on Android devices, indicating ongoing support and troubleshooting efforts for mobile deployment.
• issues/100317

• Model Compilation and Metrics Handling Issues: A reported ValueError occurs when compiling a TensorFlow model with multiple outputs using a single-entry metrics list, highlighting the inconvenience of having to specify metrics separately for each output. This points to limitations in the current metrics API design.
• issues/100319

• Security Feature Implementation Discussion: There is a discussion about implementing stored S/MIME methods to enhance Gmail message security, indicating interest in integrating advanced security protocols.
• issues/100320

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: 3

Summarized Issues:

• Build and Compatibility Errors: Several issues describe failures related to TensorFlow's compatibility with specific Python versions and system configurations. One issue highlights build failures due to SyntaxWarning errors from invalid escape sequences in Python 3.12 when compiling TensorFlow 2.10.1 with CUDA support, while another reports an ImportError caused by a DLL initialization failure in TensorFlow 2.19.0, preventing the native runtime from loading properly.
• [issues/99885, issues/100244]

• API Usage Errors: There is a reported bug involving incorrect usage of the tensorflow.keras.layers.LSTM layer in TensorFlow 2.20.0, where passing an invalid keyword argument inputs= instead of the expected input tensor leads to a TypeError about a missing required argument sequences. This indicates strict argument requirements in the API that can cause runtime errors if not followed correctly.
• [issues/100300]

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.

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

---

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: 9

Key Open Pull Requests

1. Note and Clarify the Error Raised in tf.raw_ops.ExtractGlimpse() and tf.raw_ops.ExtractGlimpseV2(): This pull request aims to improve the clarity and detail of the error messages raised by the TensorFlow operations tf.raw_ops.ExtractGlimpse() and tf.raw_ops.ExtractGlimpseV2() by updating their API definitions and source code documentation.

• URL: pull/100135

• Merged: No

• Associated Commits: 2f76b, 5cada, 6bfef, 20d77

2. Add AddressSanitizer (ASan) build support to Bazel config and macros, #99792: This pull request adds support for building TensorFlow with AddressSanitizer (ASan) using Bazel by introducing ASan-specific build flags, updating Bazel configuration files and macros to propagate these flags, and fixing related build and linker issues to enable detection of memory errors such as buffer overflows and use-after-free bugs.

• URL: pull/99900

• Merged: No

• Associated Commits: e47bf, 77010

3. tf.math.argmax support for int16 tensor in axis.: This pull request fixes a memory corruption issue in the tf.math.argmax function when the axis parameter is provided as an int16 tensor by implementing safe casting to int32 and adding support for int, int32_t, and int64_t types, along with a test case to verify correct handling of int16 inputs.

• URL: pull/99904

• Merged: No

• Associated Commits: 334ac, a30ae

Other Open Pull Requests

• Nullptr Dereference Prevention: Multiple pull requests focus on preventing nullptr dereference errors in different components of TensorFlow. These include adding missing return statements and nullptr checks to enhance code safety and stability.
• pull/99908, pull/99910

• oneDNN 3D Pooling Core Dump Fixes: Several pull requests address core dump issues in oneDNN 3D pooling operations by handling corner cases and fixing related bugs. These fixes improve the robustness of maxpooling and AvgPooling3DGrad operations.
• pull/99921, pull/99922

• Documentation and Error Message Clarifications: Some pull requests update documentation and error messages to clarify parameter behavior and improve user understanding. This includes clarifying the select_cols parameter in CSV decoding functions and error messages in checkpoint management.
• pull/100235, pull/100140

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: 1

Key Closed Pull Requests

1. This is spam: This pull request, titled "This is spam," appears to be an unmerged contribution to the TensorFlow project containing generic commit messages labeled "Summary" and a merge from a forked branch, suggesting it was likely identified as irrelevant or spam.

• URL: pull/100266

• Merged: No

• Associated Commits: 98dd5, 7c547, 82a9a

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
ILCSFNO	11	5	9	6
mihaimaruseac	0	0	0	16
psamanoelton	2	1	0	11
Venkat6871	1	0	0	11
zvoicu000	5	3	0	3
No author found	10	0	0	0
adi1220	4	1	0	5
springcrane	0	0	5	5
Aaraviitkgp	6	0	0	3
aravindhbalaji1985	0	1	2	5