- 06 Feb, 2026 1 commit
-
-
Jacket authored
Signed-off-by:Kaining Zhong <kainingz@nvidia.com>
-
- 04 Feb, 2026 1 commit
-
-
jberchtold-nvidia authored
* Update README.rst Signed-off-by:
jberchtold-nvidia <158520091+jberchtold-nvidia@users.noreply.github.com> * Update README.rst Signed-off-by:
-
- 03 Feb, 2026 3 commits
-
-
Paweł Gadziński authored
* init Signed-off-by:
Pawel Gadzinski <pgadzinski@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fixes Signed-off-by:
pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
-
Oleg Goncharov authored
* Fixed scaling-factor computation for FP32 to match the reference implementation. Signed-off-by:
Oleg Goncharov <ogoncharov@nvidia.com> * Uncommented the tuned kernel path Signed-off-by:
-
Vadim Markovtsev authored
* Support building with headers from nvidia wheels There are two changes: 1. `import nvidia` returns a namespace package with `__file__` equal to `None` 2. Add the way to force headers from nvidia wheels. Without that envvar, it's practically impossible with CUDA installed system-wide. I successfully built the package with torch using the following `uv` configuration: ``` [tool.uv.extra-build-dependencies] "transformer-engine-torch" = [ "ninja", "nvidia-cuda-crt==13.0.88", "nvidia-cuda-cccl==13.0.85", { requirement = "torch", match-runtime = true }, { requirement = "pytorch-triton", match-runtime = true }, { requirement = "nvidia-cusolver", match-runtime = true }, { requirement = "nvidia-curand", match-runtime = true }, { requirement = "nvidia-cublas", match-runtime = true }, { requirement = "nvidia-cusparse", match-runtime = true }, { requirement = "nvidia-cudnn-cu13", match-runtime = true }, { requirement = "nvidia-nvtx", match-runtime = true }, { requirement = "nvidia-cuda-nvrtc", match-runtime = true }, { requirement = "nvidia-cuda-runtime", match-runtime = true }, ] ``` Signed-off-by:Vadim Markovtsev <vadim@poolside.ai> * Apply suggestion from @ksivaman Signed-off-by:
Kirthi Shankar Sivamani <ksivamani@nvidia.com> Signed-off-by:
-
- 02 Feb, 2026 1 commit
-
-
Paweł Gadziński authored
* Code drop: Update recipes documentation and remove custom recipes from low precision training Signed-off-by:
-
- 30 Jan, 2026 1 commit
-
-
Paweł Gadziński authored
* version change Signed-off-by:
-
- 29 Jan, 2026 1 commit
-
-
Paweł Gadziński authored
fix wheel Signed-off-by:
-
- 28 Jan, 2026 2 commits
-
-
Paweł Gadziński authored
* code drop Signed-off-by:
Przemyslaw Tredak <ptrendx@gmail.com> Signed-off-by:
Paweł Gadziński <62263673+pggPL@users.noreply.github.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * changed Signed-off-by:
-
Paweł Gadziński authored
* jjit bug fix Signed-off-by:
-
- 27 Jan, 2026 1 commit
-
-
jberchtold-nvidia authored
* Use "nyu-mll/glue" instead of "glue" for encoder datasets to fix 404 error Signed-off-by:
Jeremy Berchtold <jberchtold@nvidia.com> * rename mnist dataset path Signed-off-by:
-
- 26 Jan, 2026 1 commit
-
-
Santosh Bhavani authored
* fix(examples): te_llama compatibility with HuggingFace transformers >= 4.57 The te_llama.py example was failing with HuggingFace transformers 4.57+ due to API changes in how decoder layer outputs are handled. Changes: - Handle case where hidden_states is passed as a tuple (older HF versions) - Return tensor directly instead of wrapped in tuple (HF 4.57+ expects this) - Fix regex pattern to use raw string (fixes SyntaxWarning) Error fixed: AttributeError: 'tuple' object has no attribute 'contiguous' Tested with: - transformer_engine 2.5.0 - transformers 4.57.3 - PyTorch container nvcr.io/nvidia/pytorch:25.08-py3 Signed-off-by:
Santosh Bhavani <santosh.bhavani@live.com> * docs(te_llama): add requirements.txt Signed-off-by:
-
- 25 Jan, 2026 1 commit
-
-
Tim Moon authored
* Expose option for custom op fusions Refactor fusion functions to remove index bookkeeping. Refactor fused ops to use consistent operation order. Signed-off-by:
Tim Moon <tmoon@nvidia.com> * Add tests for custom ops Signed-off-by:
-
- 23 Jan, 2026 2 commits
-
-
Oleg Goncharov authored
* Disabled the tuned NVFP4 kernels Signed-off-by:
-
Teddy Do authored
* Use correct block size for workspace in row id map creation, also shard workspace correctly based on 2nd dim of routing_map/row_id map Signed-off-by:
DoubleCheeseCheetos <hanhdp99@gmail.com> * reduce size of largest test case on single_GPU scenario to fit on L40 and A100 in CI line up Signed-off-by:
-
- 22 Jan, 2026 4 commits
-
-
Sudhakar Singh authored
* SWA (left, right) with FusedAttention changes cherry-picked from https://github.com/NVIDIA/TransformerEngine/pull/1369 Signed-off-by:
Sudhakar Singh <sudhakars@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix test_kv_cache failures Signed-off-by:
-
Chen Cui authored
* Update THD sink attention logic for newer cudnn versions THD Sink attention is supported in 9.18.0 Signed-off-by:
Chen Cui <chcui@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * update thd sink attention logic for cp>1 Signed-off-by:
Charlene Yang <8636796+cyanguwa@users.noreply.github.com>
-
Teddy Do authored
return tokens_per_experts always Signed-off-by:tdophung <tdophung@nvidia.com>
-
jberchtold-nvidia authored
* Fix cb.CUDAOptions usage for Triton 3.6.0 Signed-off-by:
-
- 21 Jan, 2026 3 commits
-
-
Przemyslaw Tredak authored
* PoC of the changes Signed-off-by:
Przemek Tredak <ptredak@nvidia.com> * Early exit from the Free function for the empty tensor Signed-off-by:
-
Oleg Goncharov authored
Signed-off-by: -
Oleg Goncharov authored
* Implemented persistent nvfp4 kernel Signed-off-by:
-
- 20 Jan, 2026 2 commits
-
-
Charlene Yang authored
* update FE to 1.17 Signed-off-by:
greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com> Signed-off-by:
-
Przemek Tredak authored
Signed-off-by:
-
- 17 Jan, 2026 1 commit
-
-
Tim Moon authored
* Add general C API for setting tensor params Signed-off-by:
Tim Moon <4406448+timmoon10@users.noreply.github.com> * Update expected error message in FP8 block-scaling test Signed-off-by:
-
- 16 Jan, 2026 1 commit
-
-
Teddy Do authored
* initial impl, not tested Signed-off-by:
-
- 15 Jan, 2026 5 commits
-
-
Jacket authored
Signed-off-by: -
jberchtold-nvidia authored
* install cmake in jax build github action Signed-off-by:
-
jberchtold-nvidia authored
disable fused attention in encoder tests for determinism Signed-off-by: -
Santosh Bhavani authored
* Move older news to Previous Signed-off-by:
-
Hongbin Liu authored
* bug fix Signed-off-by:
hongbinl <hongbinl@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Update transformer_engine/common/swizzle/swizzle_block_scaling.cu Mask to 8 bits to prevent potential bit overlap Co-authored-by:
Hongbin Liu <lhb8125@users.noreply.github.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Update transformer_engine/common/swizzle/swizzle_block_scaling.cu Co-authored-by:
-
- 14 Jan, 2026 1 commit
-
-
Teddy Do authored
* Remove pyhtorch-triton as a requirement and remove auto-fetching pytorch-triton as it is a placeeholder in pyPI Signed-off-by:
-
- 13 Jan, 2026 2 commits
-
-
Victor Oliveira authored
ONNX: Fix FP8 quantization for the second MLP in LayernormMLP Signed-off-by:Victor Oliveira <victor.oliveira@getcruise.com>
-
Paweł Gadziński authored
* code drop Signed-off-by:
-
- 10 Jan, 2026 1 commit
-
-
Tim Moon authored
Debug Doxygen and LaTeX warnings Signed-off-by:
-
- 09 Jan, 2026 2 commits
-
-
Tim Moon authored
Update list of CI users Signed-off-by: -
Kshitij Lakhani authored
* Pick a leaner set of combinations for TE JAX CP attn tests such that only those cp,dp,tp combinations are picked where cp*dp*tp is equal to num gpus Signed-off-by:
Kshitij Lakhani <klakhani@nvidia.com> * Consolidate the test cases run for different B,S,H,D and QKV layout Signed-off-by:
-
- 08 Jan, 2026 1 commit
-
-
Teddy Do authored
* Add triton version detection logic, and NVTE_USE_PYTORCH_TRITON knob for jax Signed-off-by:
-
- 07 Jan, 2026 2 commits
-
-
Teddy Do authored
* force initialization to int32 Signed-off-by:
-
Zhongbo Zhu authored
* fix Signed-off-by:
Zhongbo Zhu <zhongboz@nvidia.com> * resolve review comments Signed-off-by:
-
