- 12 Jun, 2025 3 commits
-
-
Phuong Nguyen authored
* Implemented GroupedDense and TestGroupedDense for BF16, FP16, and FP8 * Fix GroupedGemmFFI cuBLAS workspace alignment bug Signed-off-by:
Hua Huang <huah@nvidia.com> Signed-off-by:
Phuong Nguyen <phuonguyen@nvidia.com>
-
Paweł Gadziński authored
typo fix Signed-off-by:Pawel Gadzinski <pgadzinski@nvidia.com>
-
Kirthi Shankar Sivamani authored
Fix for loading old ckpt formats Signed-off-by:Kirthi Shankar Sivamani <ksivamani@nvidia.com>
-
- 10 Jun, 2025 2 commits
-
-
Kirthi Shankar Sivamani authored
* Initial basic setup Signed-off-by:
-
yuzhongw-nvidia authored
* Support MLA (qk_dim != v_dim) for AttnFuncWithCPAndKVP2P Signed-off-by:
Yuzhong Wang <yuzhongw@nvidia.com> * add UT for MLA CP Signed-off-by:
pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com> Co-authored-by:
Xin Yao <xiny@nvidia.com> Co-authored-by:
Xiaowei Ren <103958965+xrennvidia@users.noreply.github.com>
-
- 09 Jun, 2025 2 commits
-
-
Kirthi Shankar Sivamani authored
* Manage deps and add einops Signed-off-by:
-
Selvaraj Anandaraj authored
* Lora spike Signed-off-by:
Selvaraj Anandaraj <selvaraja@login-preos02.a51.clusters.nvidia.com> * Added FP8 param support Signed-off-by:
-
- 07 Jun, 2025 2 commits
-
-
Kirthi Shankar Sivamani authored
Remove extra include Signed-off-by: -
Xin Yao authored
only get skip_fp8_weight_update in fp8_graph_capturing Signed-off-by:
-
- 06 Jun, 2025 3 commits
-
-
Alp Dener authored
* added missing deallocs in Userbuffers destroyer Signed-off-by:
Alp Dener <adener@nvidia.com> * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci --------- Signed-off-by:
-
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:
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:
zhongboz <zhongboz@nvidia.com> * pass all quantizer unit tests Signed-off-by:
Keith Wyss <kwyss@nvidia.com> * add more unit test with dequantize compact supported Signed-off-by:
Tim Moon <tmoon@nvidia.com> * Remove dead logic for high-precision AGs Signed-off-by:
Tim Moon <4406448+timmoon10@users.noreply.github.com>
-
- 05 Jun, 2025 4 commits
-
-
Phuong Nguyen authored
* fix otype for fp8 gemm Signed-off-by:
-
Przemyslaw Tredak authored
* Use versioned flavor of get driver entrypoint function Signed-off-by:
Przemek Tredak <ptredak@nvidia.com> * Update the check to call the versioned API starting with CUDA 12.5 where it was added Signed-off-by:
-
jberchtold-nvidia authored
* Fix 1x quantize kernel availability check on hopper --------- Signed-off-by:Jeremy Berchtold <jberchtold@nvidia.com>
-
Kirthi Shankar Sivamani authored
* Fix NVTE_FRAMEWORK=all Signed-off-by:
-
- 03 Jun, 2025 2 commits
-
-
Charlene Yang authored
* add missing args in cross-attn Signed-off-by:
Charlene Yang <charleney@nvidia.com> * allow thd for TELayer 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 * tweak CP wording Signed-off-by:
-
jberchtold-nvidia authored
Fix NVTETensor leak in attention.cpp Signed-off-by:
-
- 02 Jun, 2025 2 commits
-
-
Kirthi Shankar Sivamani authored
minor build improvements Signed-off-by: -
jberchtold-nvidia authored
* Use 1x quantization + jax transpose on BW for performance Signed-off-by:
-
- 31 May, 2025 1 commit
-
-
Kirthi Shankar Sivamani authored
* Add warning for multi framework case Signed-off-by:
-
- 30 May, 2025 2 commits
-
-
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:
-
Tim Moon authored
Signed-off-by:
-
- 29 May, 2025 4 commits
-
-
Hua Huang authored
* Support SWA in CP Ring Attn THD striped sharding Signed-off-by:
-
Przemyslaw Tredak authored
* Changed the Tensor allocation strategy Signed-off-by:
-
Tim Moon authored
* Linear op avoids saving input tensor if weight grad is not needed Signed-off-by:
-
Phuong Nguyen authored
* removes unnecessary reshapes for FP8 GEMM * use nn.jax.scaled_matmul Signed-off-by:
-
- 28 May, 2025 3 commits
-
-
Phuong Nguyen authored
* fp8 gemm with direct quant Signed-off-by:
-
jberchtold-nvidia authored
Update jax_scaled_masked_softmax to match TE kernel implementation Signed-off-by: -
Kirthi Shankar Sivamani authored
* Fix single FW build with multi FW available Signed-off-by:
-
- 23 May, 2025 1 commit
-
-
Przemyslaw Tredak authored
* Modify the test cases Signed-off-by:
-
- 22 May, 2025 5 commits
-
-
Tim Moon authored
Activation ops support fusing backward pass with quantize Signed-off-by: -
Kirthi Shankar Sivamani authored
Document all recipes Signed-off-by: -
jberchtold-nvidia authored
Make primitive names more granular for better disabling granularity Signed-off-by: -
Kirthi Shankar Sivamani authored
* Build support for cuda 13 Signed-off-by:
-
Paweł Gadziński authored
fix saved_tensors Signed-off-by:
-
- 21 May, 2025 1 commit
-
-
Kirthi Shankar Sivamani authored
* Add missing docs for C API Signed-off-by:
-
- 20 May, 2025 3 commits
-
-
Charlene Yang authored
add docstring for CP Signed-off-by: -
Paweł Gadziński authored
* docs drop Signed-off-by:
Przemyslaw Tredak <ptrendx@gmail.com> Signed-off-by:
Paweł Gadziński <62263673+pggPL@users.noreply.github.com> * Update docs/debug/1_getting_started.rst Co-authored-by:
-
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:
-
