Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 13, 2025

• [n-gen] DO NOT repeatedly return finished child requests (#28591)

  Show description for a1d3866
  Jialin

  authored
  a1d3866

  Copy full SHA for a1d3866

• Rename clashing method names for vLLM model protocol (#27583)

  Show description for 97d1c99
  hmellor

  authored
  97d1c99

  Copy full SHA for 97d1c99

• [Docs] Add some details about what the MoE block needs for the Transformers backend (#28588)

  Show description for 3226283
  hmellor

  authored
  3226283

  Copy full SHA for 3226283

• [BugFix] Fix `mm_encoder_attn_backend` arg type checking (#28599)

  Show description for 8832fff
  njhill

  authored
  8832fff

  Copy full SHA for 8832fff

• [Bugfix] Fix SM100 gpt-oss regression due to faulty attn sink support (#28561)

  Show description for a543e67
  mgoin

  authored
  a543e67

  Copy full SHA for a543e67

• [platform] Move get_cu_count to utils (#27005)

  Show description for 2dacd57
  wangxiyuan

  authored
  2dacd57

  Copy full SHA for 2dacd57

• [ROCm][Bugfix] Revert removing setuptools version restriction (#28592)

  Show description for d75ad04
  gshtras

  authored
  d75ad04

  Copy full SHA for d75ad04

• [Docs] Update meetups.md description (#28583)

  Show description for 52eadce
  mgoin

  authored
  52eadce

  Copy full SHA for 52eadce

Commits on Nov 12, 2025

• Skip models that cannot currently init on Transformers v5 (#28471)

  Show description for 51c599f
  hmellor

  authored
  51c599f

  Copy full SHA for 51c599f

• [MoE][Kernel][Perf] Improve Shared Expert Stream Overlap (#28406)

  Show description for 69d0e90
  alexm-redhat

  authored
  69d0e90

  Copy full SHA for 69d0e90

• [Core][AMD] Migrate fully transparent sleep mode to ROCm platform (#12695)

  Show description for 4ca5cd5
  

  3 peopleauthored
  4ca5cd5

  Copy full SHA for 4ca5cd5

• [Bugfix] Adjust Marlin CUDA arch selection to 8.0+PTX;9.0+PTX (#28294)

  mgoin

  authored
  10f01d5

  Copy full SHA for 10f01d5

• [TPU] Support GCS path in VLLM_TORCH_PROFILER_DIR (#28487)

  Show description for 3eb0c26
  QiliangCui

  authored
  3eb0c26

  Copy full SHA for 3eb0c26

• [ROCM] Fix ROCm warnings, environment flag access, and GEMM kernel naming for consistency in `_aiter_ops.py` (#28464)

  Show description for d8140b9
  vllmellm

  authored
  d8140b9

  Copy full SHA for d8140b9

• [Performance][B200] Fix deepgemm prologue (#27897)

  Show description for 74a9a9f
  

  varun-sundar-rabindranath

  and

  Varun Sundar Rabindranath

  authored
  74a9a9f

  Copy full SHA for 74a9a9f

• [Misc]Fix typo in llm_engine.py (#28584)

  Show description for 478ee51
  frank-wei

  authored
  478ee51

  Copy full SHA for 478ee51

• [BugFix] Priority scheduling and spec tokens preemption (#28558)

  Show description for 58ce8d1
  andylolu2

  authored
  58ce8d1

  Copy full SHA for 58ce8d1

• [KV connector][WIP] KV cache proxy based on LMCache multi-process mode (#27902)

  Show description for 94a9ebc
  ApostaC

  authored
  94a9ebc

  Copy full SHA for 94a9ebc

• [CI] Skip "Multi-Modal Models Test (Extended) 3" test that's broken in current Transformers (#28559)

  Show description for a39dd7b
  hmellor

  authored
  a39dd7b

  Copy full SHA for a39dd7b

• [Model] [Config] Correctly identify granite-4.0-micro as non-hybrid model (#28563)

  Show description for 64d57c3
  tdoublep

  authored
  64d57c3

  Copy full SHA for 64d57c3

• [EPLB][ROCm]: support EPBL for ROCm backend (#27731)

  Show description for a1e7fa3
  

  PerryZhang01

  and

  zgplvyou

  authored
  a1e7fa3

  Copy full SHA for a1e7fa3

• Implement ARC KV cache eviction policy for CPU offloader (#27039)

  Show description for bac9045
  

  albertoperdomo2

  and

  orozery

  authored
  bac9045

  Copy full SHA for bac9045

• [Perf] Refactor cudagraph_support to enable full CUDA graphs for spec decoding with FlashInfer (#28479)

  Show description for 3044195
  benchislett

  authored
  3044195

  Copy full SHA for 3044195

• Remove deprecated fields from `CompilationConfig` (#27593)

  Show description for a742134
  hmellor

  authored
  a742134

  Copy full SHA for a742134

• [Misc] Refactor Attention kv transfer methods into decorator (#27816)

  Show description for 728a9eb
  

  NickLucche

  and

  markmc

  authored
  728a9eb

  Copy full SHA for 728a9eb

• [Refactor] Remove redundant TP gather/split in split_qkv in QwenVL (#28271)

  Show description for bc5bd45
  gcanlin

  authored
  bc5bd45

  Copy full SHA for bc5bd45

• [Performance][Hopper] Avoid M dim padding to 4x for most cases (due to cuda graphs paddings) (#28492)

  Show description for f76e85c
  alexm-redhat

  authored
  f76e85c

  Copy full SHA for f76e85c

• Fix pre-commit (and XPU) on `main` (#28556)

  Show description for 54aecd9
  hmellor

  authored
  54aecd9

  Copy full SHA for 54aecd9

• [V0 deprecation] Deprecate use_v1 parameter (#28112)

  Show description for 10138c9
  wangxiyuan

  authored
  10138c9

  Copy full SHA for 10138c9

• [Bugfix] Fix gpt_oss packed_modules_mapping (#28536)

  Show description for a9d18b5
  jeejeelee

  authored
  a9d18b5

  Copy full SHA for a9d18b5

• [ROCm] [Bugfix] Fix `fused_qknorm_rope_kernel` rocm compatibility (#28500)

  Show description for edb59a9
  tjtanaa

  authored
  edb59a9

  Copy full SHA for edb59a9

• add cpu option for p/d in nixl_connector (#28356)

  Show description for c5f10cc
  ZhengHongming888

  authored
  c5f10cc

  Copy full SHA for c5f10cc

• [KVConnector] Enable get_block_ids_with_load_errors() in LMCache connector (#27978)

  Show description for d143152
  

  3 peopleauthored
  d143152

  Copy full SHA for d143152

• [XPU]Fix crash due to removed VLLM_USE_V1 attribute (#28520)

  Show description for a4730c1
  chaojun-zhang

  authored
  a4730c1

  Copy full SHA for a4730c1

• [Model] fix glm4_moe_mtp load weights with GLM-4.6 checkpoint. (#27597)

  Show description for d3ade61
  

  wuyaoxuehun

  and

  wuao.scotty

  authored
  d3ade61

  Copy full SHA for d3ade61

Pagination

Previous

Next