- 16 May, 2025 1 commit
-
-
Selvaraj Anandaraj authored
* Added token ignoring for CE loss Signed-off-by:
Selvaraj Anandaraj <selvaraja@cw-dfw-cs-001-login-01.cm.cluster> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Added tests Signed-off-by:
root <root@cw-dfw-h100-004-210-013.cm.cluster> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Signed-off-by:
Kirthi Shankar Sivamani <ksivamani@nvidia.com>
-
- 13 May, 2025 1 commit
-
-
Charlene Yang authored
* disable sm89 and cuDNN < 9.11 for KV caching Signed-off-by:
Charlene Yang <charleney@nvidia.com> * disable some numerics tests Signed-off-by:
pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
-
- 11 May, 2025 1 commit
-
-
Kirthi Shankar Sivamani authored
* First pass refactor Signed-off-by:
-
- 07 May, 2025 1 commit
-
-
Tim Moon authored
* Initial work toward restoring UB support in te.Sequential Signed-off-by:
Tim Moon <tmoon@nvidia.com> * Forward UB linear runs, but has numerical error Signed-off-by:
-
- 06 May, 2025 1 commit
-
-
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>
-
- 05 May, 2025 1 commit
-
-
Kirthi Shankar Sivamani authored
* Move multi tensors kernels from PyTorch extensions to core 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:
-
- 30 Apr, 2025 1 commit
-
-
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:
-
- 29 Apr, 2025 1 commit
-
-
Tim Moon authored
* Check if FP8 block scaling is supported in tests Signed-off-by:
-
- 28 Apr, 2025 1 commit
-
-
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:
-
- 22 Apr, 2025 1 commit
-
-
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:
-
- 21 Apr, 2025 1 commit
-
-
Sudhakar Singh authored
* rtx5090 arch fix support Signed-off-by:
-
- 18 Apr, 2025 2 commits
-
-
Kunlun Li authored
* Add fp8_primary_weights support for blockwise scaling Signed-off-by:
kunlunl <kunlunl@nvidia.com> custom fsdp Signed-off-by:
-
Hongbin Liu authored
* split wgrad for GroupedLinear Signed-off-by:
Hongbin Liu <hongbinl@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci Signed-off-by:
-
- 15 Apr, 2025 3 commits
-
-
Li Tao authored
* support adam bf16 state Signed-off-by:
XiaobingSuper <xiaobingzhangupc@gmail.com> * use fp32 kernel but keep bf16 optimizer states to save memory 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 --------- Signed-off-by:
-
Paweł Gadziński authored
* test change Signed-off-by:
-
Paweł Gadziński authored
* fix Signed-off-by:
-
- 14 Apr, 2025 3 commits
-
-
Jianbin Chang authored
* Add fp8 weight transpose cache check in backward, and regenerated it if it does not exist Signed-off-by:
jianbinc <shjwudp@gmail.com> * Properly handle fsdp shard model weight input. Signed-off-by:
-
Xin Yao authored
* Enable MXFP8 and Per-Tensor Current Scaling for Grouped Linear Signed-off-by:
-
Autumn1998 authored
* add support for new recipe on permute_fusion, rm fp unpermute Signed-off-by:
tongliu <tongliu@nvidia.com> * fix lint Signed-off-by:
-
- 11 Apr, 2025 1 commit
-
-
Tim Moon authored
* Add option to cache activation input in FP8 Signed-off-by:
-
- 10 Apr, 2025 1 commit
-
-
kwyss-nvidia authored
* Add GEMM logic for blockwise quantized tensors. GEMM test cases included in pytorch integration. Signed-off-by:
Keith Wyss <kwyss@nvidia.com> * Update NVTE_BLOCK_SCALING for GEMM. Signed-off-by:
zhongboz <zhongboz@nvidia.com> * Edit and cleanup BF16 gather code. Signed-off-by:
Xin Yao <yaox12@outlook.com> * Update test_float8blockwisetensor.py Signed-off-by:
-
- 09 Apr, 2025 2 commits
-
-
Tim Moon authored
* Debug checkpointing with te.Sequential Signed-off-by:
-
Tim Moon authored
Explicitly specify quantized tensor usages needed for linear op backward Signed-off-by:
-
- 07 Apr, 2025 3 commits
-
-
kwyss-nvidia authored
* Add GEMM logic for blockwise quantized tensors. GEMM test cases included in pytorch integration. Signed-off-by:
-
Jianbin Chang authored
Support fp8 primary weight in fsdp training Signed-off-by:
-
Xin Yao authored
* refactor to add cp support for sbhd/bshd Signed-off-by:
-
- 04 Apr, 2025 1 commit
-
-
kwyss-nvidia authored
* Blockwise float8 quantizer and quantized tensor class. The classes are configurable for 128x128 blocksize and 1x128 blocksize via setting block_scaling_dim == 2,1 respectively. Scale tensors are stored in a format emenable for matrix multiplication, however the integration of matmul is deferred as a separate story. Fusions of quantization and DBIAS or activation functions are not yet implemented, and the dequantization is currently implemented in torch. Tests for quantization are included in C++ and pytorch layers, with exact comparison to reference quantizer behavior as well as an attempt to hit interesting branches through the API such as tensor creation in pytorch and CPP and dequantization of row and columnwise usage. Two CUDA kernels for quantization are included, and are direct ports of equivalents in the kitchen repository, where a subchannel recipe has been used for end to end training. Signed-off-by:
-
- 31 Mar, 2025 1 commit
-
-
Tim Moon authored
* Handle case where FP8 current scaling quantizer gets default process group Signed-off-by:
-
- 27 Mar, 2025 1 commit
-
-
Kirthi Shankar Sivamani authored
* Cleanup sanity tests and add CS recipe tests Signed-off-by:
-
- 25 Mar, 2025 3 commits
-
-
Charlene Yang authored
* skip cuDNN 9.8 for KV caching Signed-off-by:
Charlene Yang <8636796+cyanguwa@users.noreply.github.com> * revert from max_seqlen_kv to max_sequence_length for InferenceParams Signed-off-by:
-
Peter St. John authored
* Defer torch compilation steps until first function call Signed-off-by:
Peter St. John <pstjohn@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Fix function call in smoke test Signed-off-by:
-
Kirthi Shankar Sivamani authored
Signed-off-by:
-
- 24 Mar, 2025 1 commit
-
-
Xiaowei Ren authored
* fix dtypes of fused_attn_bwd in CP+A2A Signed-off-by:
Xiaowei Ren <xren@nvidia.com> * fix dtypes of fused_attn_bwd in CP+P2P Signed-off-by:
-
- 22 Mar, 2025 1 commit
-
-
Kunlun Li authored
* Enable fp8_primary_weights for current scaling Signed-off-by:
Kunlun Li <94586211+kunlunl@users.noreply.github.com> Signed-off-by:
-
- 18 Mar, 2025 3 commits
-
-
Przemyslaw Tredak authored
* Do not apply bias when apply_bias is False Signed-off-by:
Przemyslaw Tredak <ptrendx@gmail.com> * Fix the Gelu Aux type Signed-off-by:
-
Charlene Yang authored
* add paged attention; test_kv_cache_accuray and test_paged_attn pass Signed-off-by:
-
kwyss-nvidia authored
* Update full recompute feature to save recipe. The recompute context uses the same recipe and fp8 settings as the original fwd pass. Signed-off-by:
-
- 15 Mar, 2025 1 commit
-
-
Li Tao authored
* support tp-comm-overlap in Current Scaling recipe Signed-off-by:
-
- 14 Mar, 2025 1 commit
-
-
Kshitij Lakhani authored
* Create pytorch/dot_product_attention module and pytorch/d_p_a/utils.py Move attention logging into a separate class in pytorch/d_p_a/utils.py Signed-off-by:
-
