Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on May 26, 2026

• [ci] Add arm64 ci image (#41303)

  Show description for e19b9b1
  

  khluu

  and

  claude

  authored
  e19b9b1View commit details

  Copy full SHA for e19b9b1

  Browse repository at this point

• [Bugfix][V1] Fix TOCTOU race causing intermittent `EADDRINUSE` on multi-API-server DP startup (#42585)

  Show description for 812e7e7
  

  vadiklyutiy

  and

  gemini-code-assist[bot]

  authored
  812e7e7View commit details

  Copy full SHA for 812e7e7

  Browse repository at this point

• [KV Connector] MooncakeStore: drop dead discard_partial_chunks parameter (#43627)

  Show description for d98cbf4
  

  3 peopleauthored
  d98cbf4View commit details

  Copy full SHA for d98cbf4

  Browse repository at this point

• [Kernel] Porting fuse_minimax_qk_norm to manual fusion (#43410)

  Show description for 6e50386
  jeejeelee

  authored
  6e50386View commit details

  Copy full SHA for 6e50386

  Browse repository at this point

• [CI] Soft-fail AMD entrypoints mirror tests (#43709)

  Show description for 49b4882
  

  khluu

  and

  claude

  authored
  49b4882View commit details

  Copy full SHA for 49b4882

  Browse repository at this point

• [DSv4] Drop _get_compressed_kv_buffer in DeepseekCompressor (#43690)

  Show description for 193ce88
  WoosukKwon

  authored
  193ce88View commit details

  Copy full SHA for 193ce88

  Browse repository at this point

• [Doc] Add line limit to AGENTS.md (#43635)

  Show description for 3aea37d
  

  WoosukKwon

  and

  markmc

  authored
  3aea37dView commit details

  Copy full SHA for 3aea37d

  Browse repository at this point

• Add LM head quantization support for ModelOpt (#42124)

  Show description for 6f5b533
  meenchen

  authored
  6f5b533View commit details

  Copy full SHA for 6f5b533

  Browse repository at this point

• [ROCm] Remove MegaMoE integration in deepseek v4 (#43629)

  Show description for c8414a8
  WoosukKwon

  authored
  c8414a8View commit details

  Copy full SHA for c8414a8

  Browse repository at this point

• [MoE Refactor] W4a8 int8 oracle (#42789)

  Show description for f51bbc6
  

  bnellnm

  and

  robertgshaw2-redhat

  authored
  f51bbc6View commit details

  Copy full SHA for f51bbc6

  Browse repository at this point

• [MoE Refactor] Migrate ModelOptMxFp8FusedMoE to oracle (#42768)

  Show description for b226dda
  

  bnellnm

  and

  robertgshaw2-redhat

  authored
  b226ddaView commit details

  Copy full SHA for b226dda

  Browse repository at this point

• [Feat][DSV4] Fuse q pad into deepseek v4 fused kernel (#43162)

  zyongye

  authored
  6ab6ffbView commit details

  Copy full SHA for 6ab6ffb

  Browse repository at this point

• [ROCm][CI] Extend ROCm quick reduce coverage (#40990)

  Show description for 445ded1
  AndreasKaratzas

  authored
  445ded1View commit details

  Copy full SHA for 445ded1

  Browse repository at this point

• [Docs][ROCm] MoRI-IO Connector Usage Guide (#43603)

  Show description for d565357
  

  simondanielsson

  and

  gemini-code-assist[bot]

  authored
  d565357View commit details

  Copy full SHA for d565357

  Browse repository at this point

• Fix CuPy runtime deps and restore humming (#43530)

  Show description for a970fb5
  mmangkad

  authored
  a970fb5View commit details

  Copy full SHA for a970fb5

  Browse repository at this point

• [XPU] Fix fused MoE LoRA kernel crash on XPU by using platform-agnos num_compute_units (#43646)

  Show description for 861b977
  chaojun-zhang

  authored
  861b977View commit details

  Copy full SHA for 861b977

  Browse repository at this point

• [Model] Use AutoWeightsLoader for InternLM2 (#38278)

  Show description for ebd0692
  

  javierdejesusda

  and

  DarkLight1337

  authored
  ebd0692View commit details

  Copy full SHA for ebd0692

  Browse repository at this point

• [Reasoning] [Bugfix] Reject invalid thinking_token_budget values (#43402)

  Show description for 739af5c
  

  linzm1007

  and

  mergify[bot]

  authored
  739af5cView commit details

  Copy full SHA for 739af5c

  Browse repository at this point

• [Misc] Support interleaved custom image benchmark datasets (#43636)

  Show description for 5d09f47
  ThibaultCastells

  authored
  5d09f47View commit details

  Copy full SHA for 5d09f47

  Browse repository at this point

• [Misc][Refactor][ROCm] Convert MoRI-related envvars to extra config args (#43303)

  Show description for 681d7dd
  

  simondanielsson

  and

  tjtanaa

  authored
  681d7ddView commit details

  Copy full SHA for 681d7dd

  Browse repository at this point

• [KV Transfer] Enable HMA by default for connectors that support it (#41847)

  Show description for 755043c
  chfeng-cs

  authored
  755043cView commit details

  Copy full SHA for 755043c

  Browse repository at this point

• [Bugfix] Apply fc_norm in Eagle3DeepseekV2 combine_hidden_states (#43482)

  Show description for 97e4022
  

  yubofredwang

  and

  claude

  authored
  97e4022View commit details

  Copy full SHA for 97e4022

  Browse repository at this point

• [chores][log] change registry log from `warning` to `debug` (#43045)

  Show description for b326945
  ILikeIneine

  authored
  b326945View commit details

  Copy full SHA for b326945

  Browse repository at this point

• Add CuTe DSL sparse compressor support (#43584)

  Show description for a37e471
  

  3 peopleauthored
  a37e471View commit details

  Copy full SHA for a37e471

  Browse repository at this point

• Upgrade tpu-inference to v0.20.0 (#43394)

  CienetStingLin

  authored
  e6adbd7View commit details

  Copy full SHA for e6adbd7

  Browse repository at this point

• [CPU] Enable non-divisible GQA for decode workitems in mixed batches (#43032)

  Show description for 771e1e4
  zhejiangxiaomai

  authored
  771e1e4View commit details

  Copy full SHA for 771e1e4

  Browse repository at this point

• [GDN] GDN Prefill kernel for SM100 (#43273)

  Show description for d56612c
  gau-nernst

  authored
  d56612cView commit details

  Copy full SHA for d56612c

  Browse repository at this point

• [Bugfix][Model] Fix GPT2ForSequenceClassification sub-module prefix (#43579)

  Show description for 6f95598
  

  QingZhou-YangHY

  and

  DarkLight1337

  authored
  6f95598View commit details

  Copy full SHA for 6f95598

  Browse repository at this point

• [Frontend] Split the offline inference APIs and utils. (#43553)

  Show description for d5cf7b4
  

  noooop

  and

  gemini-code-assist[bot]

  authored
  d5cf7b4View commit details

  Copy full SHA for d5cf7b4

  Browse repository at this point

• [Bugfix] fix device mismatch in MiniCPM-o-4_5 resampler (#43194)

  Show description for f815c99
  yma11

  authored
  f815c99View commit details

  Copy full SHA for f815c99

  Browse repository at this point

• [KV Connector] Propagate MooncakeStore load failures (#42788)

  Show description for c2a4005
  Dao007forever

  authored
  c2a4005View commit details

  Copy full SHA for c2a4005

  Browse repository at this point

• [KV Connector][Bugfix] MooncakeStore: don't double-apply Eagle prune in load_mask (#43516)

  Show description for 7966fc7
  

  Dao007forever

  and

  claude

  authored
  7966fc7View commit details

  Copy full SHA for 7966fc7

  Browse repository at this point

• [DeepSeek V4] Move MegaMoE input prep kernel to nvidia/ops (#43632)

  Show description for aa2b56f
  WoosukKwon

  authored
  aa2b56fView commit details

  Copy full SHA for aa2b56f

  Browse repository at this point

• [LoRA] Add one shot triton kernel For MoE LoRA (#42290)

  Show description for ec5de7f
  jeejeelee

  authored
  ec5de7fView commit details

  Copy full SHA for ec5de7f

  Browse repository at this point

• [XPU] Ensure RNG offset alignment with PyTorch requirements in XPU sampler (#43028)

  Show description for 71d810b
  

  3 peopleauthored
  71d810bView commit details

  Copy full SHA for 71d810b

  Browse repository at this point