- 21 Oct, 2025 1 commit
-
-
Zhongbo Zhu authored
* pipeclean, fix nvfp4 padding of 32 alignment Signed-off-by:
Zhongbo Zhu <zhongboz@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * numerical test passed Signed-off-by:
Tim Moon <tmoon@nvidia.com> * Fix invalid conversion from tensor to int Signed-off-by:
pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by:
Tim Moon <4406448+timmoon10@users.noreply.github.com>
-
- 20 Oct, 2025 2 commits
-
-
Kirthi Shankar Sivamani authored
* Fix CI failures due to deterministic attention Signed-off-by:
Kirthi Shankar Sivamani <ksivamani@nvidia.com> * some more cleanup Signed-off-by:
-
fzyzcjy authored
* Update permutation.py Signed-off-by:
fzyzcjy <5236035+fzyzcjy@users.noreply.github.com> * Update permutation.py Signed-off-by:
-
- 18 Oct, 2025 1 commit
-
-
Kirthi Shankar Sivamani authored
* Support wheel build for cuda 13 Signed-off-by:
-
- 17 Oct, 2025 6 commits
-
-
Alp Dener authored
Make `CanonicalizeGemmInput()` support non-TN layout FP8 GEMM on Blackwell with column-wise/transposed data (#2233) Modified CanonicalizeGemmInput() logic to pull from column-wise data for FP8 GEMM on Blackwell when row-wise is not available. Signed-off-by:Alp Dener <adener@nvidia.com>
-
Neil Tenenholtz authored
* Fix test of FSDP2 by correcting init logic and applying autocast This fixes multiple issues in the FSDP2 test, namely 1. Previously fp8 init was performed when `args.fp8_init == False`. I have updated the logic to match what I presume was intended by leveraging the nullcontext context manager. 2. `te.fp8_autocast` was previously not called; the recipe was created but was unused. The autocast context manager now wraps the model's computation. Signed-off-by:
Neil Tenenholtz <ntenenz@users.noreply.github.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Fix typo Signed-off-by:
-
Haowen Zheng authored
Signed-off-by:
将来 <jianglai.zhw@alibaba-inc.com> Co-authored-by:
-
Tim Geypens authored
Signed-off-by:Tim Geypens <tim.geypens@gmail.com>
-
Kevin Tong authored
* CUDA RHT Signed-off-by:
Kevin Tong <kevin@augmentcode.com> * Fix cuda graphs Signed-off-by:
-
Kshitij Lakhani authored
* Fix imports in test for deprecated jax.experimental.pjit Signed-off-by:
Kshitij Lakhani <klakhani@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Fix: Pass NamedSharding instead of PartitionSpec to compare_ops() so that when the in and out sharding is used to create a jitted function, it has the mesh info Signed-off-by:
Kshitij Janardan Lakhani <klakhani@login-eos01.eos.clusters.nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Signed-off-by:
-
- 16 Oct, 2025 3 commits
-
-
Przemek Tredak authored
Signed-off-by:Przemek Tredak <ptredak@nvidia.com>
-
xiaoxi-wangfj authored
* [PyTorch] Add record_stream and untyped_storage func op in QuantizedTensor Signed-off-by:
xiaoxi-wangfj <690912414@qq.com> * Update transformer_engine/pytorch/tensor/float8_blockwise_tensor.py Co-authored-by:
-
Selvaraj Anandaraj authored
Signed-off-by:
Selvaraj Anandaraj <selvaraja@login-ptyche02.ptyche.clusters.nvidia.com> Co-authored-by:
-
- 15 Oct, 2025 2 commits
-
-
Santosh Bhavani authored
* Enhance Latest News section with recent TE and FP8 developments - Adds NVFP4 pretraining research paper with PR #2177 reference Signed-off-by:
Santosh Bhavani <santosh.bhavani@live.com> * update nvfp4 reference Signed-off-by:
-
Paul Gibbons authored
* fixes for start_end_list usage in TE debug Signed-off-by:
Paul Gibbons <pgibbons@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Signed-off-by:
Paweł Gadziński <62263673+pggPL@users.noreply.github.com>
-
- 14 Oct, 2025 5 commits
-
-
Tim Moon authored
* Require cuDNN 9.14.0+ for fused attention with FP8 current scaling Signed-off-by:
-
Kirthi Shankar Sivamani authored
* Initial API change Signed-off-by:
-
Kshitij Lakhani authored
* Add BRCM support when creating a test mask for fused attn Signed-off-by:
-
Evgeny Tsykunov authored
* Fix update_quantized in ref nvfp4 quantizer Signed-off-by:
Evgeny <etsykunov@nvidia.com> * Subclass quantization API Signed-off-by:
-
Kshitij Lakhani authored
Fix test path so that it gets triggered Signed-off-by:
-
- 13 Oct, 2025 5 commits
-
-
Selvaraj Anandaraj authored
* FSDP grad fusion support Signed-off-by:
Selvaraj Anandaraj <selvaraja@login-ptyche01.ptyche.clusters.nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Re-factored grad overwriting usage Signed-off-by:
Selvaraj Anandaraj <selvaraja@nvidia.com> * Update transformer_engine/pytorch/ops/fused/backward_linear_add.py Co-authored-by:
-
jberchtold-nvidia authored
assertion check Signed-off-by:Jeremy Berchtold <jberchtold@nvidia.com>
-
jberchtold-nvidia authored
* Improve error message for cublas fp8 gemm with incorrect shape Signed-off-by:
-
Peter St. John authored
Signed-off-by:
Peter St. John <pstjohn@nvidia.com> Co-authored-by:
-
Selvaraj Anandaraj authored
* Added multi-layout support for attention Signed-off-by:
Pawel Gadzinski <pgadzinski@nvidia.com> --------- Signed-off-by:
-
- 09 Oct, 2025 5 commits
-
-
Peter St. John authored
Don't pickle an empty dict in LayerNorm and BasicOperation layers Signed-off-by:
-
jberchtold-nvidia authored
Signed-off-by:
Phuong Nguyen <phuonguyen@nvidia.com>
-
Kirthi Shankar Sivamani authored
* Update minimum python version to 3.10 and update CI Signed-off-by:
-
Kirthi Shankar Sivamani authored
Deprecate old float8_tensor.py Signed-off-by: -
Kirthi Shankar Sivamani authored
Catch unsupported GEMM during recipe init Signed-off-by:
-
- 08 Oct, 2025 2 commits
-
-
Hua Huang authored
* Try async copy of grouped GEMM group_sizes data Signed-off-by:
Hua Huang <huah@nvidia.com> --------- Signed-off-by:
-
Kirthi Shankar Sivamani authored
Unblock path for fusing NVFP4 quantize and bgrad Signed-off-by:
-
- 07 Oct, 2025 3 commits
-
-
vcherepanov-nv authored
Signed-off-by:Vladimir Cherepanov <vcherepanov@nvidia.com>
-
Kirthi Shankar Sivamani authored
* Improve docstring for NVFP4 recipe Signed-off-by:
Przemyslaw Tredak <ptrendx@gmail.com> Signed-off-by:
Copilot <175728472+Copilot@users.noreply.github.com> Signed-off-by:
-
Phuong Nguyen authored
* reuse amax for current scaling Signed-off-by:
-
- 06 Oct, 2025 3 commits
-
-
Kirthi Shankar Sivamani authored
* Update test requirements for HF Signed-off-by:
-
Kiv Chen authored
[Build] fix: python platlib path Signed-off-by:
Kiv Chen <sdckivenchen@gmail.com> Co-authored-by:
-
Phuong Nguyen authored
* not fuse bias for output all reduction case + unit tests Signed-off-by:
-
- 05 Oct, 2025 1 commit
-
-
Przemyslaw Tredak authored
* Added the NVFP4 part to the low precision tutorial Signed-off-by:
-
- 04 Oct, 2025 1 commit
-
-
Kirthi Shankar Sivamani authored
* Fix in FP8 attention selection logic Signed-off-by:
-
