All pull requests

Pull requests

• Guard fused MoE auxiliary loss integer offsets

  community-contribution
  #3569·arcusbuilds opened Sep 24, 2026·

  ·
  1

• Add DSv4 sparse attention bindings to TE

  #3566·sudhakarsingh27 opened Sep 23, 2026Member·

  ·

• [common] Reject mismatched bias/out dtype for non-FP8/FP4 GEMM

  community-contribution
  #3565·dundysm opened Sep 23, 2026·

  ·
  1

  10

• Update intialize_ub docstring

  community-contribution
  #3563·Prachi-kushwaha opened Sep 23, 2026·

  ·
  1

  3

  

• Bump NCCL EP submodule commit

  #3561·phu0ngng opened Sep 22, 2026Collaborator·

  ·
  2

• [Commo] Optimize grouped MXFP8 requantization

  #3560·Oleg-Goncharov opened Sep 22, 2026Collaborator·

  ·
  9

• [Common] Enable EP with FP32/FP16 payload dtypes

  2.21
  #3558·phu0ngng opened Sep 21, 2026Collaborator·

  ·
  11

• Treat total_num_pages as a physical KV cache budget, not the page-table size

  attentioncommunity-contribution
  #3557·0z5a opened Sep 21, 2026·

  ·
  2

• Add an opt-in batch-invariant BF16 GEMM

  community-contribution
  #3556·0z5a opened Sep 21, 2026·

  ·
  4

• Add an opt-in float32 reduction dtype for the row-parallel all-reduce

  community-contribution
  #3555·0z5a opened Sep 21, 2026·

  ·
  5

• Reuse the host split sizes GroupedLinear's forward already computed

  community-contribution
  #3554·0z5a opened Sep 21, 2026·

  ·
  3

  

• Isolate FlashAttention 2/3 imports so a broken install cannot break TE

  attentioncommunity-contribution
  #3553·0z5a opened Sep 21, 2026·

  ·
  1

  2

• Accumulate delayed wgrads into shared (tied) parameters

  community-contribution
  #3552·0z5a opened Sep 21, 2026·

  ·
  1

  26

• Support learnable softmax (attention sink) with cp_comm_type="p2p"

  attentioncommunity-contribution
  #3551·0z5a opened Sep 20, 2026·

  ·
  5

• KDA attention

  2.21attention
  #3548·ksivaman opened Sep 18, 2026Member·

  ·
  4

• [PyTorch] Compile Linear and Bias operations with forward fusion

  #3547·pggPL opened Sep 18, 2026Collaborator·

  ·

• [PyTorch] Enable LayerNormLinear UB AG overlap under no_grad

  community-contribution
  #3546·ravimajeti opened Sep 18, 2026·

  ·
  1

  1

• [PyTorch] Keep DistributedWeight objects on ctx across saved-tensor hooks

  community-contribution
  #3545·xrennvidia opened Sep 18, 2026Collaborator·

  ·
  1

  4

• feat: vmm slot for local cuda graph offload

  community-contribution
  #3544·Xdydy opened Sep 18, 2026·

  ·
  6

• Use PyTorch CUTLASS for BF16 grouped GEMM on SM100

  community-contribution
  #3542·wujingyue opened Sep 18, 2026Contributor·

  ·
  9

• [PyTorch][CP] Support compact metadata with FA4

  #3540·sudhakarsingh27 opened Sep 17, 2026Member·

  ·
  4

• Prototype of green context + VMM localization

  community-contribution
  #3537·WanZzzzzz opened Sep 17, 2026Contributor·

  ·

• test(attention): anchor the context-parallel suite to an independent reference

  attentioncommunity-contribution
  #3536·nvegesna-netizen opened Sep 17, 2026Contributor·

  ·
  2

• [Docs] Refresh README

  #3535·sbhavani opened Sep 17, 2026Collaborator·

  ·
  1

• [PyTorch] [torch.compile] torch.compile support for LayerNormLinear and LayerNormMLP

  #3534·pggPL opened Sep 17, 2026Collaborator·

  ·
  2