Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit History

Commits on Feb 12, 2026

• Add config file for fused MoE for Nemotron (TP4, B200) (#34411)

  Show description for dea6351
  danisereb

  authored
  dea6351

  Copy full SHA for dea6351

• [ROCm] Enable MXFP4 MoE weight pre-shuffling on gfx950 and update aiter (#34192)

  Show description for 8a798be
  

  4 peopleauthored
  8a798be

  Copy full SHA for 8a798be

• [V0 Deprecation] Remove code related to per-request logits processors (#34400)

  Show description for fb455ed
  DarkLight1337

  authored
  fb455ed

  Copy full SHA for fb455ed

• Fix Mistral config remap to accept compressed-tensors quantization #34028 (#34104)

  Show description for f589761
  baonudesifeizhai

  authored
  f589761

  Copy full SHA for f589761

• [bugfix] refactor FunASR's _get_data_parser (#34397)

  Show description for 386bfe5
  

  AllenDou

  and

  zixiao

  authored
  386bfe5

  Copy full SHA for 386bfe5

• [Bugfix] Fix Sparse24 Compressed Tensors models (#33446)

  Show description for e9cd691
  

  kylesayrs

  and

  mgoin

  authored
  e9cd691

  Copy full SHA for e9cd691

• Fix DeepSeek-OCR tensor validation for all size variants (#34085)

  Show description for 80f2ba6
  

  yichuan-w

  and

  cursoragent

  authored
  80f2ba6

  Copy full SHA for 80f2ba6

• [BugFix] Fix DP chunking (#34379)

  Show description for 136b0bf
  

  LucasWilkinson

  and

  bnellnm

  authored
  136b0bf

  Copy full SHA for 136b0bf

• [Refactor] Pass Renderer to Input Processor (#34329)

  Show description for b96f731
  DarkLight1337

  authored
  b96f731

  Copy full SHA for b96f731

• [Refactor] Move validation to params definitions (#34362)

  Show description for ced2a92
  DarkLight1337

  authored
  ced2a92

  Copy full SHA for ced2a92

• [Bug Fix] Fix `naive_block_assignment` always defaulting to False due to arg misalignment (#33848)

  Show description for e1d97c3
  RunkaiTao

  authored
  e1d97c3

  Copy full SHA for e1d97c3

• [Bugfix] Fix MTP accuracy for GLM-5 (#34385)

  Show description for ec12d39
  mgoin

  authored
  ec12d39

  Copy full SHA for ec12d39

• [Refactor] Replace `activation: str` with `MoEActivation` enum (#33843)

  Show description for ff1f83b
  mgoin

  authored
  ff1f83b

  Copy full SHA for ff1f83b

Commits on Feb 11, 2026

• [Bugfix] Fix some issues with MoERunner PR #32344 (#34371)

  Show description for 31d992d
  bnellnm

  authored
  31d992d

  Copy full SHA for 31d992d

• [Bugfix] Fix more multimodal tests for transformers V5 (#34334)

  Show description for 527ca32
  zucchini-nlp

  authored
  527ca32

  Copy full SHA for 527ca32

• [Bugfix] send None sentinel on final commit so server properly sends transcription.done (#33963)

  Show description for 5458eb8
  

  pjs102793

  and

  njhill

  authored
  5458eb8

  Copy full SHA for 5458eb8

• [Benchmarks] Reduce ready checker log verbosity (#34349)

  Show description for 144d9b7
  tomasruizt

  authored
  144d9b7

  Copy full SHA for 144d9b7

• [GPT-OSS] Remove unnecessary contiguous (#34337)

  Show description for 83e26c8
  elvischenv

  authored
  83e26c8

  Copy full SHA for 83e26c8

• [Bugfix] Enable attn quantization of Llama-4 by correctly permuting scales for rope (int8, fp8) (#34243)

  Show description for 11c7ace
  

  eldarkurtic

  and

  Your Name

  authored
  11c7ace

  Copy full SHA for 11c7ace

• [Bugfix] fix default is_neox_style is True for deepseek (#34353)

  Show description for be7f3d5
  xyDong0223

  authored
  be7f3d5

  Copy full SHA for be7f3d5

• [Multimodal] Expose `mm_processor_kwargs` for `DummyInputsBuilder` (#34330)

  Show description for 0ab0610
  Isotr0py

  authored
  0ab0610

  Copy full SHA for 0ab0610

• [Model Runner V2] Init cuda graph pool when necessary (#33217)

  Show description for ffb3d55
  xinyu-intel

  authored
  ffb3d55

  Copy full SHA for ffb3d55

• [Docs] Fix typo ("defult") and double spacing (#34348)

  Show description for 48134a2
  

  3 peopleauthored
  48134a2

  Copy full SHA for 48134a2

• [ROCm] [aiter] Split KV cache update for AiterFlashAttention (#33681)

  Show description for 64f570a
  kliuae

  authored
  64f570a

  Copy full SHA for 64f570a

• Reapply [Attention][FA3] Update FA3 to include new swizzle optimization (#34043)

  Show description for c7914d3
  LucasWilkinson

  authored
  c7914d3

  Copy full SHA for c7914d3

• Responses harmony system message structured (#34268)

  Show description for 1b87565
  Kimahriman

  authored
  1b87565

  Copy full SHA for 1b87565

• [NVIDIA][test] Tests for flashinfer TRTLLM BF16 MoE (#33715)

  Show description for 275e0d2
  

  Linda-Stadter

  and

  pavanimajety

  authored
  275e0d2

  Copy full SHA for 275e0d2

• Make JAIS compatible with Transformers v5 (#34264)

  Show description for 0f5e55e
  hmellor

  authored
  0f5e55e

  Copy full SHA for 0f5e55e

• Make Qwen3VL compatible with Transformers v5 (#34262)

  Show description for 1e9204b
  

  hmellor

  and

  ywang96

  authored
  1e9204b

  Copy full SHA for 1e9204b

• [Bugfix][CPU] Fix llama4 inference on CPU (#34321)

  Show description for 05339a7
  bigPYJ1151

  authored
  05339a7

  Copy full SHA for 05339a7

• [Docs] Reduce time spent generating API docs (#34255)

  Show description for 40b8f55
  hmellor

  authored
  40b8f55

  Copy full SHA for 40b8f55

• [Frontend] Exploit tokenizers "new stream" in FastIncrementalDetokenizer (#34217)

  Show description for e09546c
  njhill

  authored
  e09546c

  Copy full SHA for e09546c

• [torch.compile] Enable AR+rms fusion by default available for `-O2` (#34299)

  Show description for addac0e
  ProExpertProg

  authored
  addac0e

  Copy full SHA for addac0e

• [Chore] Move `BaseRenderer` to `base.py` (#34308)

  Show description for 675a22e
  DarkLight1337

  authored
  675a22e

  Copy full SHA for 675a22e

• [XPU][9/N] clean up existing ipex code/doc (#34111)

  Show description for cb9574e
  jikunshang

  authored
  cb9574e

  Copy full SHA for cb9574e