vllm-project/vllm

Pull Requests Commits

Merge branch 'main' into codex/moe-masked-activation

mgoin committed 13 hours ago

Verified 34ecf7ca

[Test] Add ROCm AITER FP8 MLA prefill accuracy test (#51457)

aarushjain29 committed 14 hours ago

Verified 7581c56c

[Core] Make the GPU sync check thread-local and fix its suppressors (#51455)

njhill committed 14 hours ago

Verified fbff187d

[Perf] Avoid some more unnecessary GPU<->CPU syncs (#51458)

njhill committed 14 hours ago

Verified 9b0afeb4

[Bugfix][MRV2] Reserve spec-decode lookahead blocks in V2 warmup (#51438)

njhill committed 14 hours ago

Verified d608dfab

[Perf][Sparse MLA] Drop the atomic contention in the index remap (#50365)

njhill committed 15 hours ago

Verified 9e6be4a7

[Bugfix][Parser] Confirm reasoning end when an Inkling content block opens (#49876)

Vegetog committed 16 hours ago

Verified 1c1077c6

[Bugfix][Parser] Prevent Inkling block-end leakage with tools (#51391)

taking-lying-flat committed 22 hours ago

Verified 75231eff

[BugFix] Preserve divergent FA hits with external Mamba state (#51468)

majunze2001 committed 1 day ago

Verified 58d3918e

[Bugfix] Fix LFM2 ShortConv prefix breaking quant ignore list (#51495)

xiaopusun committed 1 day ago

Verified 0fe9a916

Add torch compile for qwen3_vl encoder (#40116)

gty111 committed 1 day ago

Verified 653ebb52

[ROCm][CI] Baseline legacy extensions in the Torch ABI audit (#50805)

AndreasKaratzas committed 1 day ago

Verified 643c125f

[Bugfix][MM] Avoid device sync in FusedInputNorm initialization (#51435)

AndreasKaratzas committed 1 day ago

Verified 6e189011

[CI] Refresh hybrid Model Runner V2 coverage (#51410)

ZJY0516 committed 1 day ago

Verified 44351f81

[CPU] Optimize routed FP8/MXFP4 MoE GEMM dispatch (#50949)

tianmu-li committed 1 day ago

Verified 700d39b5

[Bugfix][Multimodal] Fix PyNvVideoCodec video backend returning NCHW instead of NHWC (#51076)

davidjpyu committed 1 day ago

Verified d3621c1e

[BugFix] Reject invalid data-parallel RPC ports (#51469)

aoshen02 committed 1 day ago

Verified 22a17592

[Bugfix][models_multimodal] Remote HF python code misses importing class (#51427)

gchinora committed 1 day ago

Verified d21fed95

[Bugfix][multi_modal] Fix pos_ids being unitialized for minicpmv2.6 in hf runner (#51432)

music-dino committed 1 day ago

Verified a828536f

[CI] Guard remote-code Transformers compatibility (#51451)

AndreasKaratzas committed 1 day ago

Verified 12da9b23

[Bugfix][Quantization] Fix INT8 W8A8 MoE crash in TritonExperts (#51411)

djramic committed 1 day ago

Verified 5eaa70da

[Bugfix] Fix ZMQ port TOCTOU race in shm_broadcast MessageQueue (#50960)

aoshen02 committed 1 day ago

Verified f7ef489e

[Kimi][MM] disable kimi_vit's dynamic torch.compile for TPU (#51196)

lk-chen committed 1 day ago

Verified 7f58e829

[Bugfix] Close usage telemetry HTTP sessions (#51219)

matteso1 committed 1 day ago

Verified 1c008a34

[Perf] Optimize DeepSeek V3.2 sequence parallelism (#51434)

WoosukKwon committed 1 day ago

Verified e644c8cd

Fix Humming DeepEP v2 test setup

mgoin committed 1 day ago

804b91a7

[CI Test] Add specific unit test for mrv2 offloading (#51440)

yewentao256 committed 1 day ago

Verified c39076fe

[Perf] Improve `--linear-backend` filtering (#48735)

askliar committed 1 day ago

Verified a801e71c

[Perf] Narrow DeepSeek V3.2 eager CUDA graph region (#51425)

WoosukKwon committed 1 day ago

Verified 46b58640

[Spec Decode] Register Qwen3.6 dSpark acceptance coverage (#51310)

mgoin committed 1 day ago

Verified c8fe1d57

Older