All pull requests

Pull requests

• Fix archive links across documentation publishing locations

  community-contribution
  #3665·emcallister20 opened Oct 9, 2026·4 of 7 tasks·

  ·
  2

• [PyTorch] Fix NVFP4 weight views of the single-group fused grouped MLP

  #3662·sraman-rgb opened Oct 9, 2026Collaborator·13 tasks·

  ·
  1

• [Common][PyTorch] Add row-scaled NVFP4 support to grouped-tensor GEMM

  community-contribution
  #3661·cael-ling opened Oct 9, 2026Contributor·2 of 13 tasks·

  ·
  10

• GroupedTensor Machinery for MXFP8 Dispatch input passed from Megatron

  #3660·vthumbe1503 opened Oct 9, 2026Collaborator·13 tasks·

  ·
  3

• Fix FP8 current scaling FakeTensor output shape

  community-contribution
  #3659·sanjana658 opened Oct 9, 2026·

  ·
  3

• [Common] Swizzle MXFP8 scaling factors with one CTA per scale tile

  #3658·sraman-rgb opened Oct 8, 2026Collaborator·13 tasks·

  ·
  7

  

• [PyTorch] Distributed tests set device in NCCL init and manually shut down NCCL

  testing
  #3655·timmoon10 opened Oct 8, 2026Member·5 of 13 tasks·

  ·
  8

• Preserve router probability precision in fused grouped MLP

  community-contribution
  #3653·mkhona-nvidia opened Oct 8, 2026·13 tasks·

  ·
  1

• [PyTorch] Support FP64 second moment in FusedAdam

  community-contribution
  #3652·deepakn94 opened Oct 8, 2026Contributor·5 of 13 tasks·

  ·
  2

• Guard zero-gradient blocks in the default FP8 block-scaling recipe

  community-contribution
  #3651·HollowMan6 opened Oct 8, 2026Contributor·8 of 13 tasks·

  ·
  4

• Fuse the residual and bias adds into their neighbours: proj bias in the MXFP8 GEMM epilogue (with BGRAD in the dgrad epi

  community-contribution
  #3650·denys-fridman opened Oct 8, 2026·

  ·

• [Common][PyTorch] Add NVFP4 row-scaled grouped cast with graph-safe path

  community-contribution
  #3649·cael-ling opened Oct 8, 2026Contributor·2 of 13 tasks·

  ·
  5

• [PyTorch] Add BF16 fused grouped MLP support

  community-contribution
  #3646·tingyangk opened Oct 7, 2026Contributor·7 of 12 tasks·

  ·

• perf: Improvements to nvfp4 4over6 cuda kernel

  community-contribution
  #3643·CarlosGomes98 opened Oct 7, 2026Contributor·13 tasks·

  ·
  2

• [Common] More CuTeDSL MXFP8 specialized kernels

  #3642·kainzhong opened Oct 7, 2026Collaborator·13 tasks·

  ·

• [PyTorch] Pass canonical layouts to cuDNN grouped GEMM in fused grouped MLP

  org-contribution
  #3641·hwanseoc opened Oct 6, 2026Member·8 of 13 tasks·

  ·
  3

• [Common][JAX] Reconstruct EpHandle via ncclEpImportHandle

  #3639·phu0ngng opened Oct 6, 2026Collaborator·13 tasks·

  ·
  10

• [Draft][JAX] MoE block optimization

  #3638·jberchtold-nvidia opened Oct 6, 2026Collaborator·1 of 13 tasks·

  ·
  5

• cuBLASMp NVFP4 support

  community-contribution
  #3634·almogsegal opened Oct 6, 2026Contributor·13 tasks·

  ·
  5

• [PyTorch][Common] Add MXFP8 and NVFP4 QDQ for mixed-precision GEMM

  #3633·negvet opened Oct 6, 2026Collaborator·1 of 13 tasks·

  ·
  3

• [Common] Make Grouped MXFP8 quantization concurrency safe for varying last cases

  #3632·kainzhong opened Oct 5, 2026Collaborator·7 of 13 tasks·

  ·
  1

  2

• [PyTorch] Add NVTE_CUTEDSL_FUSED_GROUPED_MLP_WARN_FALLBACK to warn on grouped MLP fusion fallback

  community-contribution
  #3631·ghadiaravi13 opened Oct 5, 2026Contributor·6 of 13 tasks·

  ·
  8

• Reuse cuBLASMp's all-gathered grad_output in the wgrad GEMM

  community-contribution
  #3629·almogsegal opened Oct 5, 2026Contributor·1 of 13 tasks·

  ·
  2

  

• Fix NVFP4 compile metadata for non-matrix inputs

  #3628·pggPL opened Oct 5, 2026Collaborator·7 of 13 tasks·

  ·
  1

• Let the CUDA allocator manage cuBLAS workspaces

  community-contribution
  #3625·Vtmpas opened Oct 5, 2026·8 of 13 tasks·

  ·
  14