Weekly GitHub Report for Tensorflow: September 01, 2025 - September 08, 2025 (12:06:14)

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 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. RTX 5000 series support (CUDA_ERROR_INVALID_HANDLE): This issue reports a bug where TensorFlow does not natively support NVIDIA RTX 5000 series GPUs (compute capability 12.0), resulting in CUDA errors such as CUDA_ERROR_INVALID_HANDLE when attempting to run GPU operations outside of NVIDIA’s NGC containers. The user seeks clarification on whether native support for these GPUs is planned or in progress, as the current workaround using containers is inconvenient and not aligned with official TensorFlow documentation.

   • The comments confirm the problem remains unresolved, with multiple users experiencing similar CUDA errors on RTX 5000 GPUs using TensorFlow 2.20 and CUDA 12.9. Workarounds involving NVIDIA containers are acknowledged as the only reliable solution currently, though users express frustration with the lack of native support and hope for an official fix or timeline for inclusion.
   • Number of comments this week: 5

2. Missing type annotations: This issue reports that some type annotations are missing in TensorFlow, causing Pyright to raise errors when accessing the numpy method on tf.Variable objects. The user highlights that this problem occurs even with the PyPI version of TensorFlow and requests that maintainers address the missing type annotations centrally.

   • The comments include a detailed workaround involving adding type annotations directly in the TensorFlow source and creating custom type stubs to resolve Pyright errors locally, but the original reporter clarifies that the issue persists with the official PyPI package, emphasizing the need for an upstream fix by TensorFlow maintainers.
   • Number of comments this week: 3

3. tf.clip_by_value produces inconsistent results for -0.0 between CPU and GPU: This issue reports a bug where the TensorFlow function tf.clip_by_value produces inconsistent results for the input value -0.0 when run on CPU versus GPU, with the CPU returning -0.0 and the GPU returning 0.0. This discrepancy, although mathematically equivalent, may lead to unexpected behavior in applications requiring consistent outputs across different hardware devices.

   • The comments reveal that this inconsistency is not isolated to tf.clip_by_value but also occurs in other TensorFlow APIs such as tf.math.maximum, tf.math.minimum, tf.nn.relu, and tf.round, where CPU returns -0.0 and GPU returns 0.0; one commenter acknowledged the issue and indicated they would investigate it when possible.
   • Number of comments this week: 2

4. tf.image.combined_non_max_suppression aborts process on wrong-rank inputs (should raise Python exception instead): This issue reports that the function tf.image.combined_non_max_suppression causes a fatal C++ abort when given inputs with incorrect tensor ranks, instead of raising a Python exception as expected. The user provides a minimal reproducible example demonstrating that supplying boxes and scores tensors with lower-than-expected ranks leads to a process crash rather than a catchable error.

   • A commenter confirmed the issue occurs in TensorFlow versions 2.19.0 and 2.20.0 and shared a Colab gist to reproduce the problem, supporting the original report without additional discussion.
   • Number of comments this week: 1

5. Wheels for 2.20.0 on tensorflow.org: This issue concerns the absence of TensorFlow 2.20.0 wheels on the official tensorflow.org installation page, which currently does not support Python 3.13. The user is inquiring about the expected availability of these new wheels to enable installation for the latest Python version.

   • The comment agrees that the installation links should be updated and notes that the links have already been fixed in a related pull request on the NixOS nixpkgs repository.
   • 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 an NVIDIA RTX 3050 Ti GPU with the appropriate CUDA 12.4 and NVIDIA driver 535 installed. The user reports difficulties with driver compatibility and has tried reinstalling TensorFlow multiple times, but continues to receive a warning that TensorRT could not be found, impacting their ability to use TensorFlow with TensorRT optimizations.
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 calling 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 triggers a SystemError and OverflowError. The problem occurs in eager execution mode, as demonstrated by a specific example where a shape tensor with a large uint64 value causes the built-in function isinstance to fail internally, indicating an underlying error in handling such large shape values.
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 comprehensive 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 highlights a discrepancy in the TensorFlow documentation for the Round operation, where it states that complex tensors are supported as input, but in practice, attempting to use a complex tensor with this operation results in an error. The user reports that they must manually apply the Round operation to the real and imaginary parts separately, indicating that the current implementation does not handle complex inputs as documented, which suggests a documentation bug needing correction.
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 fatal check failure error "Check failed: d < dims()" when invoked with certain inputs. The problem occurs on Linux Ubuntu 20.04.3 LTS using Python 3.11.8, and the user has provided a minimal reproducible example demonstrating that the operation crashes due to an invalid dimension check in the tensor shape handling code.

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: 11

Summarized Issues:

• Type Annotations Missing in TensorFlow API: The TensorFlow tf.Variable class lacks type annotations for the numpy method in version 2.19.1, causing static type checkers like Pyright to raise errors. Users request that these annotations be added centrally to improve type checking and developer experience.
• issues/99580

• GPU Support and CUDA Errors: TensorFlow 2.20 does not natively support NVIDIA RTX 5000 series GPUs (compute capability 12.0) on Linux with CUDA 12.9, resulting in CUDA_ERROR_INVALID_HANDLE errors during GPU kernel launches. Additionally, running certain Keras models with spatial self-attention causes GPU crashes with CUDA_ERROR_ILLEGAL_ADDRESS and cuDNN fallback errors, indicating issues with GPU kernel execution and error handling.
• issues/99592, issues/99631

• Build and Packaging Failures: Building TensorFlow 2.20.0 encounters multiple issues including linker errors and AddressSanitizer container-overflow errors on Ubuntu 22.04 with ASan enabled, as well as a FileNotFoundError due to missing numpy core include directories during packaging. These build problems hinder successful compilation and require guidance on proper build configurations.
• issues/99792, issues/99835

• Runtime Import and Symbol Errors: Importing TensorFlow on Windows 10 with Python 3.13 fails due to a DLL initialization error in the internal module _pywrap_tensorflow_internal. Similarly, building a Python program with tflite-runtime using Buildozer for Android results in an 'undefined symbol' error related to TensorFlow Lite's builtin operator registration, indicating runtime and linking issues.
• issues/99859, issues/99605

• TensorFlow Wheel Availability and Installation Issues: The absence of TensorFlow 2.20.0 wheel packages on tensorflow.org prevents installation and Python 3.13 support via pip on Linux systems, blocking users from upgrading or using the latest Python version with TensorFlow.
• issues/99639

• Inconsistent Behavior in TensorFlow Functions: The function tf.clip_by_value produces inconsistent results for the input -0.0 between CPU and GPU devices, with the CPU preserving the negative zero sign and the GPU returning positive zero. This discrepancy can lead to unexpected behavior in computations relying on sign preservation.
• issues/99759

• Opaque Internal Errors During Model Training: Training an LSTM model on an RTX 5070Ti GPU using NVIDIA's NGC container fails with an internal graph execution error related to the CudnnRNN operation, without providing a useful error message. This lack of diagnostic information complicates troubleshooting and resolution.
• issues/99870

• Error Handling and Exception Reporting: The TensorFlow function tf.image.combined_non_max_suppression causes a fatal C++ abort when given inputs with incorrect tensor ranks, instead of raising a Python exception. This behavior prevents graceful error handling and proper debugging by users.
• issues/99583

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: 6

Summarized Issues:

• SparseTensor support in math operations: TensorFlow's math operations like tf.asinh currently do not support SparseTensor inputs, causing a ValueError when such inputs are used. Users request either direct support for SparseTensor inputs or clearer documentation that conversion to dense tensors is required.
• issues/98994

• TensorFlow Lite GPU delegate compatibility: The MobileViT model converted to TensorFlow Lite fails to run on a Pixel 4 device using the TFLite GPU delegate due to incompatibility with dynamic-sized tensors in the model graph. This results in execution errors when using the GPU delegate.
• issues/99047

• TFLite concatenation node activation fusion: Users face difficulty obtaining a TensorFlow Lite concatenation node with a fused activation function, as the code enforces no activation on concatenation nodes during TFLite conversion. This raises questions about whether such fusion is unsupported.
• issues/99450

• TensorFlow import mutex lock error on macOS ARM: TensorFlow 2.20.0 on macOS ARM CPUs causes a mutex lock blocking error during module import, triggered by conflicts with PyArrow imports. Temporary workarounds include downgrading to TensorFlow 2.19.0 or adjusting import order and dependencies.
• issues/99464

• Documentation issues in tf.nn.pool(): The docstring for the dilations parameter in tf.nn.pool() is incorrect when padding is "SAME", failing to state that pooling with dilation rates greater than one is unsupported. Additionally, outdated reference links need updating for clarity.
• issues/99517

• TensorFlow Lite native library page alignment: TensorFlow Lite version 2.17.0 does not compile with the new 16 KB native library page alignment required by Google Play, which replaces the previous 4 KB alignment. This causes build failures under the updated alignment requirement.
• issues/99644

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: 5

Key Open Pull Requests

1. tf.image: raise ValueError for wrong-rank inputs in combined_non_max_suppression: This pull request adds explicit Python-level rank validation for the boxes and scores inputs in tf.image.combined_non_max_suppression, raising a clear ValueError for incorrect input ranks to prevent segmentation faults and improve error handling, along with adding a regression test and updating build files accordingly.

• URL: pull/99646

• Merged: No

• Associated Commits: eca51, 4f44e, 3caa7, 076a7

2. Fix crash when running TF Serving to handle parallel requests: This pull request addresses a crash in TensorFlow Serving caused by shared variables in the mkl_fused_batch_norm_op during parallel requests by moving these variables into the Compute function to provide each client thread with its own local copies, thereby eliminating the race condition.

• URL: pull/99852

• Merged: No

• Associated Commits: 9c235, 68265, 9a3d2

3. Add type annotation to numpy method in BaseResourceVariable (Fixes #99580): This pull request adds a type annotation to the numpy method in TensorFlow’s BaseResourceVariable class to specify that it returns a np.ndarray, thereby resolving the static type checking issue reported in issue #99580.

• URL: pull/99585

• Merged: No

• Associated Commits: f851c, abbcb

Other Open Pull Requests

• Support for Blackwell: This pull request introduces support for the Blackwell platform in TensorFlow, expanding the range of hardware compatibility. The addition is clearly indicated by both the title and the commit message, ensuring that the framework can leverage Blackwell's capabilities.
• pull/99751

• Consistency in tf.clip_by_value behavior: This pull request fixes an inconsistency in the tf.clip_by_value function by making sure that clipping -0.0 on a GPU returns positive zero (0.0), aligning GPU behavior with CPU behavior. This change ensures consistent results across different devices, improving reliability for users.
• pull/99780

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: 7

Key Closed Pull Requests

1. Add gcc-toolset-13: This pull request proposes adding gcc-toolset-13 to the project, specifically updating the Dockerfile.rocm.manylinux_2_28 to include this toolset as part of addressing the issue SWDEV-553729.

• URL: pull/99844

• Merged: No

• Associated Commits: f5f98, b5400, 1537a, cece9, 7ae16, e624e, 09335, 2211b, 2a929, 6c915, 07cab, 4d444, 575fd, 67b5d, a2bad, 468cf, 7ccb1, 58fc0, db5ae, 9e21f, da24c, 3515d, 8cf48, 09b52, ccced, a382c, 532df, b854f, 58d80, f4770, 3c370, e5dd7, 9f051, 1afb6, f854d, 554f8, 68c43, 2baac, 33a41, e554c, 1c5d7, 4d453, d7f29, 4ca78, 3ac1e, ed3e4, 5788e, 75503, 04c3e, e2acb, 188d1, 1d388, 87930, c1375, 66ca7, b0d33, 713b6, 09053, b6971, d9163, 76f24, 252a9, 7ec4c, 2464a, 8aedd, 071ee, 3a90e, 67d8f, ac4cb, 02686, 25619, e60b7, 70057, fd54d, b74d5, 72a9f, d71ec, f22b6, 8df41, 22772, b1e81, 34e1c, b45da, f8452, 03daf, 68e1e, e1d97, b3932, a5407, 28861, 79765, 467f7, cf3c4, 0e049, aeabf, 4c21b, 03b3d, 62c0d, b664f, 9019a, 496f7, 0091d, d5f0c, 16c82, ae453, 9de27, eec13, 15ab9, 4bc26, a3f8d, 34a99, 89a93, 7b820, a80fd, 656ad, 5063b, 21b23, 409ac, 74fa1, 94825, e5b0d, 63041, 08a73, 4f6e7, 0d805, 7311c, 3609a, a957d, 36f2e, 7491f, 662a9, bcb99, 4805c, 6d183, 879c8, 3baf7, f27a1, b2772, 71018, a1ae8, 74f23, 53140, 2fff6, ea13c, 46f72, 959a7, 255c7, 100ae, f6025, 9614a, d1df2, 847f7, 6caa5, e2592, 32ed3, 1e913, 10b74, 5260f, 6f0af, 584ea, a7600, 84679, ce67b, 00d7c, b51de, 5747e, 3cdde, bb559, 79eb0, 3dc94, 44403, 8c039, 5a61b, b5189, 18ba8, b6854, d91b5, 1db8a, c599e, 59d28, 4f530, 9041c, 6f8ea, 34634, ae46a, 368fd, cd72e, 6ba8c, 0106c, 5247c, 98922, 891a5, 3271f, b2568, 4653c, 6f548, f591c, 0c288, 92981, ab7b7, 1bb11, eaa2a, c0191, 18455, e4619, 936d4, d02ca, 372df, 5f4be, b9691, a8fe5, c0a85, 79b5e, 5a4d2, ae214, bbb7d, ef82a, 43397, 5b221, de4cc, 58d26, ea85f, 89106, 2cbf4, 2272f, b2eeb, a80fb, bf589, 72fbb, bb311, 95349, 54533, 54182, 0b52b, 7eb3a, 268ec, bda1b, 2e17d, 24cab, be6ea, 43503, 791f3, 2a65f, bfec6, bf807, 3d2b3, 9243f, 789ed, 54518, 1a615

2. Develop upstream qa rocm71: This pull request attempts to develop and synchronize upstream support for ROCm 7.1 by removing unnecessary forward declarations, updating GPU architecture configurations, passing AMDGPU targets to build tools, removing GCC compiler support, adjusting testing scripts, and disabling or fixing failing tests related to ROCm integration in TensorFlow, but it was not merged.

• URL: pull/99769

• Merged: No

• Associated Commits: 7c927, 5bdb8, 6e502, 509a6, 61aee, 1bb97, 85937, 1c5ec, 10850, 2c716, b1e73, 80949, 60d7d, d25d2, edde8, 859c6, d48f0, 292c9, 722e7, 46e74, a1b2a, b5fd7, 1f9d9, afd80, e66c8, 30130, 97ed2, b922a, 5d4c5, 50d60, 1f963, 84e49, 7ff66, 45fe4, 959ca, 4f789, 88b0f, 491f0, fc0c8

3. Add SparseTensor support for math operations : This pull request adds and tests dispatcher support for several unary math operations (asinh, sin, tan, asin, atan, atanh) on TensorFlow's SparseTensor data structure, enabling these functions to operate correctly on sparse inputs.

• URL: pull/99322

• Merged: 2025-09-04T14:48:32Z

• Associated Commits: 1ffcb, 0b9e3, 22666

Other Closed Pull Requests

• GPU kernel memory alignment fixes: This pull request addresses misaligned shared memory accesses in GPU concat and split kernels by aligning the shared memory buffer to the stricter alignment requirement between the data type T and IntType, with a minimum alignment of 16 bytes. This ensures stability and performance improvements on newer GPU architectures such as NVIDIA's Blackwell series.
• pull/99046

• Documentation correction for dilations parameter: This pull request fixes the docstring of the dilations parameter specifically when the padding is set to "SAME", resolving issue #99517. The update clarifies the behavior of the parameter in this padding mode to improve user understanding.
• pull/99518

• Java tensor example addition: This pull request adds example code in Java demonstrating how to create and manipulate a 2x3 float32 tensor using the Tensor class. It includes instructions on populating the tensor with decimal values and accessing its elements in a flattened order.
• pull/99601

• Unmerged commit: This pull request refers to a commit made to the TensorFlow repository that was not merged. No further details or changes were incorporated from this pull request.
• pull/99862

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
psamanoelton	2	1	0	12
mmakevic-amd	14	0	0	0
springcrane	0	0	8	6
ILCSFNO	4	3	3	3
mihaimaruseac	0	0	0	12
Aaraviitkgp	7	0	0	3
Venkat6871	1	0	0	9
No author found	10	0	0	0
adi1220	4	1	0	5
zvoicu000	3	2	0	3