- 28 Sep, 2025 1 commit
-
-
dongchl authored
-
- 24 Sep, 2025 1 commit
-
-
dongcl authored
-
- 19 Sep, 2025 2 commits
- 18 Sep, 2025 1 commit
-
-
yuguo authored
-
- 28 Aug, 2025 1 commit
-
-
yuguo authored
-
- 26 Aug, 2025 1 commit
-
-
yuguo authored
-
- 25 Aug, 2025 1 commit
-
-
yuguo authored
-
- 19 Aug, 2025 1 commit
-
-
evt_fugx1 authored
-
- 14 Aug, 2025 1 commit
-
-
Kshiteej K authored
* fix: update grad_output quant to avoid redundant work Signed-off-by:
kshitij12345 <kshitijkalambarkar@gmail.com> * add test Signed-off-by:
Kirthi Shankar Sivamani <ksivamani@nvidia.com> --------- Signed-off-by:
-
- 09 Aug, 2025 1 commit
-
-
Daniel Stokes authored
* fix: Add stream synchronization before destroying MPI communicator (#1979) Signed-off-by:
djns99 <40156487+djns99@users.noreply.github.com> * feat: Implement column-wise userbuffer overlap for comm+GEMM operations Add support for overlapping column-wise allgather communication with GEMM operations to improve training performance: * **Core infrastructure changes:** - Update bulk_overlap_columnwise_ag() to accept explicit stream parameter - Modify userbuffers send/recv loops to use rank-ordered iteration - Add userbuffers_send_all/recv_all function declarations * **Python integration:** - Add bulk_overlap_ag_with_external_gemm() C++ extension function - Expose new overlap function via pybind11 bindings - Update overlap method configurations to include more ring_exchange ops * **LayerNorm MLP optimization:** - Enable column-wise quantization for FC2 gradient output - Implement overlap of allgather communication with FC2 DGRAD GEMM - Use fill_userbuffers_buffer_for_all_gather for efficient buffering This optimization allows overlapping communication and computation phases more effectively, reducing training wall-clock time by hiding allgather latency behind GEMM execution. Signed-off-by:
Tim Moon <4406448+timmoon10@users.noreply.github.com>
-
- 08 Aug, 2025 1 commit
-
-
Paweł Gadziński authored
* turn on userbuffers for layers without debug Signed-off-by:
Pawel Gadzinski <pgadzinski@nvidia.com> * code drop Signed-off-by:
pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by:
-
- 06 Aug, 2025 2 commits
-
-
hx authored
[PyTorch] fix input_quantizer usage for save_original_input; fix blockwise FP8 convert_and_update_tensor (#1978) * fix input_quantizer in save_original_input bwd Signed-off-by:
Hongxiao Bai <hongxiaob@nvidia.com> * fix get shape of blockwise tensor with only compact colwise data Signed-off-by:
-
yuguo authored
-
- 01 Aug, 2025 1 commit
-
-
Hongbin Liu authored
* disable wgrad accumulation and reduce in backward() And manually launch it in backward_dw() Signed-off-by:
Hongbin Liu <hongbinl@nvidia.com> * format Signed-off-by:
-
- 23 Jul, 2025 1 commit
-
-
Charlene Yang authored
* fix current device for cuDNN/cuBLAS handles Signed-off-by:
Charlene Yang <8636796+cyanguwa@users.noreply.github.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * add unit test Signed-off-by:
-
- 22 Jul, 2025 1 commit
-
-
Tim Moon authored
* Debug linear layer when saving original input and using debug quantizer Signed-off-by:
Tim Moon <tmoon@nvidia.com> * Workaround bugs with quantizing with only column-wise usage Signed-off-by:
-
- 21 Jul, 2025 1 commit
-
-
Tim Moon authored
Reset FP8 weight workspace if usages are invalid Signed-off-by:
-
- 17 Jul, 2025 1 commit
-
-
hx authored
* save original input Signed-off-by:
Xin Yao <xiny@nvidia.com> Co-authored-by:
-
- 16 Jul, 2025 1 commit
-
-
Paweł Gadziński authored
* some initial code Signed-off-by:
Paweł Gadziński <62263673+pggPL@users.noreply.github.com> * fix Signed-off-by:
root <root@prenyx0221.a51.clusters.nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * lint fix Signed-off-by:
Pawel Gadzinski <pgadzinski@gmail.com> * Update build.yml Signed-off-by:
-
- 12 Jul, 2025 1 commit
-
-
Jan Bielak authored
* Fix clearing tensor data in backward removing is_first_op Signed-off-by:
Jan Bielak <jbielak@nvidia.com> * Misc fixes Signed-off-by:
-
- 09 Jul, 2025 1 commit
-
-
Xin Yao authored
* Fix align_size Signed-off-by:
-
- 01 Jul, 2025 2 commits
-
-
Selvaraj Anandaraj authored
* Added MCore fsdp support for TE 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 * Refactored based on new MCore FSDP Signed-off-by:
-
guyueh1 authored
Fix layernorm output shape in LayernormLinear, make it align with input shape Signed-off-by:
Guyue Huang <guyueh@nvidia.com> Co-authored-by:
-
- 26 Jun, 2025 2 commits
-
-
xiaoxi-wangfj authored
* [PyTorch|common] Implement unpadding kernel for FP8 1. Add multi-tensor unpadding kernel 2. Replace split+cat with unpadding kernel in Fp8Padding and Fp8Unpadding 3. Add unpadding with padding unit tests Signed-off-by:
xiaoxi-wangfj <690912414@qq.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * add license Signed-off-by:
-
Zhongbo Zhu authored
* finish python ref impl for bulk alloc Signed-off-by:
zhongboz <zhongboz@nvidia.com> * c++ bulk alloc worked, still draft version Signed-off-by:
-
- 13 Jun, 2025 2 commits
-
-
Tim Moon authored
* Do not initialize quantized weights with column-wise usage in inference mode Signed-off-by:
-
Daniel Stokes authored
* Add support for overlapping wgrad NCCL AG with dgrad GEMM Signed-off-by:
dastokes <dastokes@dastokes-dvt-01.nvidia.com>
-
- 12 Jun, 2025 2 commits
-
-
Paweł Gadziński authored
typo fix Signed-off-by: -
Kirthi Shankar Sivamani authored
Fix for loading old ckpt formats Signed-off-by:
-
- 11 Jun, 2025 1 commit
-
-
yuguo authored
-
- 07 Jun, 2025 1 commit
-
-
Xin Yao authored
only get skip_fp8_weight_update in fp8_graph_capturing Signed-off-by:
-
- 06 Jun, 2025 2 commits
-
-
Phuong Nguyen authored
* refactor the multi_stream utils + implement nvte_multi_tensor_quantize in TE/Common * implement GroupedQuantizer and grouped_quantize in jaxx * fix logical_axes_names for transpose tensor in ScaledTensor Signed-off-by:
Phuong Nguyen <phuonguyen@nvidia.com> --------- Signed-off-by:
Hua Huang <huah@nvidia.com> Co-authored-by:
Ming Huang <mingh@nvidia.com>
-
Zhongbo Zhu authored
[PyTorch] FP8 Subchannel Recipe With FP8 Gather And Configurable Scaling Factor Tensor Swizzling (#1707) * functional kernel for columnwise + no-transpose option, still hacky Signed-off-by:
Keith Wyss <kwyss@nvidia.com> * add more unit test with dequantize compact supported Signed-off-by:
-
- 30 May, 2025 1 commit
-
-
Evgeny Tsykunov authored
* Quantizer update Signed-off-by:
Evgeny Tsykunov <etsykunov@etsykunov-mlt.client.nvidia.com> * Update import Signed-off-by:
Evgeny <etsykunov@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Introduce _update_weight_quantizers and _get_weight_tensors/_get_weight_quantizers Signed-off-by:
Przemyslaw Tredak <ptredak@nvidia.com>
-
- 27 May, 2025 2 commits
- 20 May, 2025 3 commits
-
-
Peter St. John authored
* Use an empty torch tensor to indicate no fp8 information in extra_state Signed-off-by:
Peter St. John <pstjohn@nvidia.com> * Add huggingface from_pretrained / save_pretrained tests Adds integration tests to ensure models containing TransformerLayer objects can be saved and loaded using the from_pretrained and save_pretrained methods. Signed-off-by:
-
yuguo authored
-
yuguo authored
-
