vllm-project/vllm

Pull Requests Commits

[CI] Shard Kernels Quantization Test to 6 NVIDIA shards, pin AMD mirror at 2

khluu committed 2 hours ago

64d3c973

[Attention][DSA] Take the native decode path for MTP=3 on SM90 (#52164)

zobinHuang committed 5 hours ago

Verified 63a9a501

[Bugfix][Helm] Fix chart resource references (#51664)

iwannagotobed committed 6 hours ago

Verified aa310035

[Doc] Update model support information (#52289)

jeejeelee committed 6 hours ago

Verified 69e0e58d

[Bugfix] Fix Cosmos3-Edge processor after transformers 5.15 release (#51989)

bastefaniak committed 6 hours ago

Verified 20405bfb

[MRV2][Multimodal] Enable encoder cuda graph for model runner v2 (#49852)

Isotr0py committed 6 hours ago

Verified 66728feb

[Frontend] Consolidate entrypoint exception handler (#52261)

noooop committed 6 hours ago

Verified b8165e5e

[Perf][Frontend] Vectorize Cohere binary embedding bit-packing (#52277)

fangchenli committed 6 hours ago

Verified 103c419f

[XPU]bump up vllm_xpu_kernels to 0.1.13.2 (#52138)

jikunshang committed 7 hours ago

Verified 624999aa

[PP][XPU]Overlap async-scheduling PP sampled-token broadcast with compute (#51650)

yisustc committed 7 hours ago

Verified 3c8676ae

[CI] Increase extended generation test timeout (#52252)

LucasWilkinson committed 7 hours ago

Verified d4c24e6f

[CPU] Fold the MXFP4 block scale in 2 instructions instead of 4 (#51583)

ccaadaro committed 7 hours ago

Verified bda4c3e8

[XPU][CI/Release][3/N] Add xpu wheel release to release pipeline (#52108)

jikunshang committed 10 hours ago

Verified 8e6d8e4f

[Bugfix][CPU][RISC-V] Fix build: make FP32Vec copy constructors non-explicit (#51099)

velonica0 committed 11 hours ago

Verified ac7509e2

Add Muse Glimmer model support (#51655)

xianbaoqian committed 11 hours ago

Verified 6adad087

Detect ROCm wheel variant from environment for precompiled wheels. (#49365)

aarushjain29 committed 11 hours ago

Verified 59d1af56

[Bugfix][NIXL] Include transfer mode (push/pull) in the compatibility hash (#50620)

tzulingk committed 11 hours ago

Verified 653cc6fa

[Bugfix][Refactor] Keep Qwen3Next layer boundaries sequence parallel (#50685)

kzwrime committed 12 hours ago

Verified c05d75aa

[Kernel][Perf] Add fused CUDA post-conv MTP decode kernel for Qwen3.5 GDN (#51674)

Jie-Fang committed 12 hours ago

Verified 1be36283

[UT] fix device of test_outputs.py (#52237)

mayuyuace committed 12 hours ago

Verified d18bb7b3

[Kernel] Add B12X dense linear backends (#52016)

lukealonso committed 13 hours ago

Verified 3c79b1a8

[XPU] [Bugfix] process ragged weights in xpu linear backend (#52118)

zufangzhu committed 14 hours ago

Verified fe4c5dcd

[Bugfix] Reject negative token ids as out-of-vocabulary (#51795)

ECMGit committed 14 hours ago

Verified b216db3e

[Kernel] Gemma-4 FA4 FP8 Kernel (#48666)

jhaotingc committed 15 hours ago

Verified 827a2af8

[Bugfix] Reject NUL byte in structured_outputs.regex (#51796)

ECMGit committed 15 hours ago

Verified 38f097fa

[Model Runner V2][Spec Decode] Add KV cache support for multi-layer MTP (#50062)

TheEpicDolphin committed 16 hours ago

Verified f80b66f5

[Bugfix] Fix .../mrope.py::apply_interleaved_rope() when torch.compile is used in torch==2.13 (#52005)

bastefaniak committed 17 hours ago

Verified 71b0da7c

[Bugfix] Reapply 50869 (#52223)

benchislett committed 17 hours ago

Verified 51def784

[Bugfix] Correct prompt lengths for timed_traces benchmark (#45423)

s3woz committed 18 hours ago

Verified 63550515

[Attention] Fix FlashInfer SM12x prefill with sinks (#52148)

askliar committed 18 hours ago

Verified b652dedd

Older