Commits

Branch selector

main

User selector

All users

All time

Commit history

Commits on Mar 20, 2026

• Fix DP coordinator ZMQ port TOCTOU (#37452)

  Show description for ca1ac1a
  itayalroy

  authored
  ca1ac1a

  Copy full SHA for ca1ac1a

• [ROCm][Bugfix] fix cache block size mismatch for aiter unified attention (#37606)

  Show description for 4ca3fa6
  divakar-amd

  authored
  4ca3fa6

  Copy full SHA for 4ca3fa6

Commits on Mar 19, 2026

• [Bugfix] Fix Deepseekv32 tool parser when stream interval > 1 (#36056)

  sfeng33

  authored
  be12afd

  Copy full SHA for be12afd

• [Bug] Fix EmbedIOprocessor "classify" <-> "embed" (#37573)

  Show description for df3c029
  yewentao256

  authored
  df3c029

  Copy full SHA for df3c029

• [Refactor] Remove dead code in pooling model (#37572)

  Show description for 2be1a0f
  yewentao256

  authored
  2be1a0f

  Copy full SHA for 2be1a0f

• Fix AttributeError in Qwen3.5 GDN layers with quantized models (#37448)

  Show description for 4120a05
  

  3 peopleauthored
  4120a05

  Copy full SHA for 4120a05

• [CI][BugFix][AMD] Don't set VLLM_ROCM_USE_AITER anymore in test_rocm_aiter_topk since its not necessary (#36996)

  Show description for 98ff042
  rasmith

  authored
  98ff042

  Copy full SHA for 98ff042

• [Performance] Enable Triton autotuning disk cache by default (#37188)

  Show description for b55156e
  arpera

  authored
  b55156e

  Copy full SHA for b55156e

• test Qwen/Qwen3-4B-Instruct-2507 for unbacked (#36064)

  Show description for 112944f
  laithsakka

  authored
  112944f

  Copy full SHA for 112944f

• [MoE Refactor] Rename "naive" all2all backend (#36294)

  Show description for 91be5f9
  bnellnm

  authored
  91be5f9

  Copy full SHA for 91be5f9

• Comment fix for async rl example (#35244)

  Show description for 4ee847e
  hao-aaron

  authored
  4ee847e

  Copy full SHA for 4ee847e

• [ROCm][CI] Cleaning and restructuring amd-ci legacy pipeline (#34839)

  Show description for 040a505
  AndreasKaratzas

  authored
  040a505

  Copy full SHA for 040a505

• [MoE Refactor] DefaultMoERunner simplifcation (#33049)

  Show description for 9279c59
  bnellnm

  authored
  9279c59

  Copy full SHA for 9279c59

• [Log] Log once in local node by default (#37568)

  Show description for 7454096
  yewentao256

  authored
  7454096

  Copy full SHA for 7454096

• [CI] Add retry with 4x backoff to HTTP fetches for transient failures (#37218)

  Show description for fb8b5e0
  AndreasKaratzas

  authored
  fb8b5e0

  Copy full SHA for fb8b5e0

• Fix `SpeculatorsConfig` now that `PreTrainedConfig` is a `dataclass` in Transformers (#37574)

  Show description for e5d96dc
  hmellor

  authored
  e5d96dc

  Copy full SHA for e5d96dc

• [Bugfix] Fix AttributeError when serving MXFP8 models with DeepGEMM installed (#37358)

  Show description for daa05bf
  

  EdalatiAli

  and

  gemini-code-assist[bot]

  authored
  daa05bf

  Copy full SHA for daa05bf

• [torch.compile][BE][Multimodal] Remove requirement to set_model_tag to avoid cache conflict (#37345)

  Show description for 7769b58
  Lucaskabela

  authored
  7769b58

  Copy full SHA for 7769b58

• [P/D] AnthropicMessages add kv_transfer_params for PD disaggregation (#37535)

  Show description for 2f9f946
  chaunceyjiang

  authored
  2f9f946

  Copy full SHA for 2f9f946

• [CPU][UX] Do not crash when tcmalloc/libiomp are not ldpreloaded (#37561)

  Show description for 2890aec
  fadara01

  authored
  2890aec

  Copy full SHA for 2890aec

• [CI] Gate pre-commit on `ready` label or number of contributions (#37544)

  Show description for 34f093b
  hmellor

  authored
  34f093b

  Copy full SHA for 34f093b

• Run MacOS smoke test on daily `cron` job instead of every commit (#37567)

  Show description for 4dce832
  hmellor

  authored
  4dce832

  Copy full SHA for 4dce832

• [Misc] Cleanup more configs and processors (#37560)

  Show description for 657855a
  DarkLight1337

  authored
  657855a

  Copy full SHA for 657855a

• [Bug] Fix fp8 trtllm MoE modular kernel supported routing methods (#37346)

  Show description for e27b8ba
  wzhao18

  authored
  e27b8ba

  Copy full SHA for e27b8ba

• [MRV2] Use fp32 for draft logits (#37526)

  Show description for 40b8363
  WoosukKwon

  authored
  40b8363

  Copy full SHA for 40b8363

• [1/n] Migrate permute_cols to libtorch stable ABI (#31509)

  Show description for 8b10e4f
  mikaylagawarecki

  authored
  8b10e4f

  Copy full SHA for 8b10e4f

• Remove deprecated reasoning_content message field(part-2) (#37480)

  Show description for 104605c
  

  12 peopleauthored
  104605c

  Copy full SHA for 104605c

• [LoRA] Minor improvements to LoRA log (#37557)

  Show description for 96266f1
  

  jeejeelee

  and

  DarkLight1337

  authored
  96266f1

  Copy full SHA for 96266f1

• Cap the number of API servers to 1 when using Elastic EP. (#37466)

  Show description for 7c0cf3b
  SageMoore

  authored
  7c0cf3b

  Copy full SHA for 7c0cf3b

• Stop bench CLI from recursively casting all configs to `dict` (#37559)

  Show description for 572b432
  hmellor

  authored
  572b432

  Copy full SHA for 572b432

• [Misc] Clean up processing logic (#37541)

  Show description for 9515c20
  DarkLight1337

  authored
  9515c20

  Copy full SHA for 9515c20

• [Bugfix] Add Kimi-K2.5 reasoning/tool parser aliases and tool_call_id support (#37438)

  Show description for c63ca2b
  DorBernsohn

  authored
  c63ca2b

  Copy full SHA for c63ca2b

• [CI] Merge `cleanup_pr_body.yml` and `reminder_comment.yml` (#37552)

  Show description for a32eaf5
  hmellor

  authored
  a32eaf5

  Copy full SHA for a32eaf5

• Fix KV Offloading + MLA AssertionError by using num_kv_heads=1 in cpu… (#37536)

  Show description for e390742
  

  xueliangyang-oeuler

  and

  546827391

  authored
  e390742

  Copy full SHA for e390742

• [Model] Remove unnecessary `get_language_model` (#37545)

  Show description for 7a6ebcb
  DarkLight1337

  authored
  7a6ebcb

  Copy full SHA for 7a6ebcb