"git@developer.sourcefind.cn:kecinstone/2024-pra-vllm.git" did not exist on "1b20639a43e811f4469e3cfa543cf280d0d76265"
- 09 May, 2025 1 commit
-
-
Tim Moon authored
* Avoid spurious warning with non-FP8 GroupedLinear Signed-off-by:
Tim Moon <tmoon@nvidia.com> * Use `QuantizedTensorBase` Signed-off-by:
Kirthi Shankar Sivamani <ksivamani@nvidia.com> Signed-off-by:
-
- 08 May, 2025 3 commits
-
-
Paweł Gadziński authored
* features drop Signed-off-by:
Pawel Gadzinski <pgadzinski@nvidia.com> * Update transformer_engine/debug/features/utils/stats_computation.py Co-authored-by:
Przemyslaw Tredak <ptrendx@gmail.com> Signed-off-by:
Paweł Gadziński <62263673+pggPL@users.noreply.github.com> * Update transformer_engine/debug/features/disable_fp8_layer.py Co-authored-by:
pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
-
Li Tao authored
* use lru to cache torch.Tensor() Signed-off-by:
lit <lit@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * remove duplicated definition Signed-off-by:
-
Xiaowei Ren authored
Signed-off-by:
Xiaowei Ren <xren@nvidia.com> Co-authored-by:
-
- 07 May, 2025 5 commits
-
-
Tim Moon authored
* Initial work toward restoring UB support in te.Sequential Signed-off-by:
-
Kirthi Shankar Sivamani authored
Add build isolation to workflow Signed-off-by: -
jberchtold-nvidia authored
Signed-off-by:Jeremy Berchtold <jberchtold@nvidia.com>
-
Peter St. John authored
Signed-off-by:
Peter St. John <pstjohn@nvidia.com> Co-authored-by:
-
Santosh Bhavani authored
* added a direct link to the quickstart notebook right after the code examples section Signed-off-by:
Santosh Bhavani <sbhavani@nvidia.com>
-
- 06 May, 2025 2 commits
-
-
Przemyslaw Tredak authored
* Changes to Linear Signed-off-by:
Przemek Tredak <ptredak@nvidia.com> * Fix Signed-off-by:
Tim Moon <4406448+timmoon10@users.noreply.github.com>
-
jberchtold-nvidia authored
* Fix L2 test_custom_call_compute.py L2 tests Signed-off-by:
-
- 05 May, 2025 4 commits
-
-
Phuong Nguyen authored
* removes unneccessary reshapes for FP8 GEMM * use nn.jax.scaled_matmul Signed-off-by:
Phuong Nguyen <phuonguyen@nvidia.com> --------- Signed-off-by:
-
Kirthi Shankar Sivamani authored
Signed-off-by: -
Kirthi Shankar Sivamani authored
* Move multi tensors kernels from PyTorch extensions to core Signed-off-by:
-
jberchtold-nvidia authored
* Enforce input sharding of norm primitive does not shard hidden dim Signed-off-by:
-
- 03 May, 2025 1 commit
-
-
Xin Yao authored
* Fix autocast deprecation warnings Signed-off-by:
Xin Yao <xiny@nvidia.com> * merge main Signed-off-by:
-
- 02 May, 2025 1 commit
-
-
Kirthi Shankar Sivamani authored
* Update README.rst Signed-off-by:
-
- 01 May, 2025 3 commits
-
-
Tim Moon authored
Debug UB conversion function from NVTEShape to std::vector Signed-off-by: -
Tim Moon authored
Fix bug when reshaping norm output from LayerNormLinear Signed-off-by:
-
Phuong Nguyen authored
* exclude GroupedGemm APIs Signed-off-by:
-
- 30 Apr, 2025 3 commits
-
-
jberchtold-nvidia authored
Fix distributed layernorm test failure Signed-off-by: -
Paweł Gadziński authored
* init Signed-off-by:
-
Kirthi Shankar Sivamani authored
* Move FP8 block scaling kernels to core Signed-off-by:
-
- 29 Apr, 2025 6 commits
-
-
kwyss-nvidia authored
* Reapply "Allow NVTEShape to own data." (#1703) This reverts commit 91405eb4 . Signed-off-by:
Keith Wyss <kwyss@nvidia.com> * Update code so that data is replaced by an array. Signed-off-by:
-
Xiaowei Ren authored
* fix recompilation of out and lse correction in p2p+bshd/sbhd Signed-off-by:
-
Xiaowei Ren authored
* build pybind of sm_arch in TE-Pytorch Signed-off-by:
-
Tim Moon authored
* Check if FP8 block scaling is supported in tests Signed-off-by:
-
HuangHuang authored
Prevent precision issues by making get_old_fp8_meta_tensors_for_recompute() return copies of amaxes/scales rather than references to originals. Signed-off-by:
HuangHuang <huang.huang@mthreads.com> Co-authored-by:
-
jberchtold-nvidia authored
* Update test_helper.py and add QuantizeConfig class for CurrentScaling Signed-off-by:
-
- 28 Apr, 2025 4 commits
-
-
Kirthi Shankar Sivamani authored
* Add support for nvidia cu* lib wheels Signed-off-by:
-
Kunlun Li authored
* Prevent using fp8 weights + non-fp8 computation Signed-off-by:
kunlunl <kunlunl@nvidia.com> * Using warnings instead of raising an error Signed-off-by:
-
Alp Dener authored
[PyTorch] Fix cuBLAS workspace leak in applications that initialize+destroy Userbuffers more than once (#1715) safeguarded cuBLAS workspace expansion in initialize_ub() to avoid exponential growth across repeat initializations Signed-off-by:
Alp Dener <adener@nvidia.com> Co-authored-by:
-
Kshitij Lakhani authored
* Move MultiHeadAttention into its own file. Modify tests and files in t_e/pytorch to import from the new MHA module Signed-off-by:
Kshitij Janardan Lakhani <klakhani@nvidia.com> * Resolving lost MHA changes from PR 1614 as a result of rebase Signed-off-by:
-
- 25 Apr, 2025 2 commits
-
-
Nicolas Castet authored
Fixes #1692 Signed-off-by:Nicolas Castet <26874160+nvcastet@users.noreply.github.com>
-
Kirthi Shankar Sivamani authored
Update FSDP example instructions Signed-off-by:
-
- 24 Apr, 2025 2 commits
-
-
jberchtold-nvidia authored
Introduce nvte_memset to provide a fill kernel that is faster than cudaMemsetAsync for small sizes (#1716) * nvte_memset fills single float value Signed-off-by:
-
Kirthi Shankar Sivamani authored
Signed-off-by:
-
- 22 Apr, 2025 3 commits
-
-
Kirthi Shankar Sivamani authored
* Move radix sort to core Signed-off-by:
-
Sudhakar Singh authored
* add support for `sb1d` freqs tensor in Fused RoPE Signed-off-by:
Sudhakar Singh <sudhakars@nvidia.com> * add `start_positions` variable to `apply_rotary_pos_emb` function to make staggered rope application faster Signed-off-by:
-
jberchtold-nvidia authored
* [JAX-Q] Single GPU current scaling for JAX Signed-off-by:
-
