vllm-project/vllm

Pull Requests Commits

remove torch 2.9, 2.10 workround

yewentao256 committed 38 days ago

77230471

[CI/Build] Skip Prithvi/Terratorch model-registry tests when terratorch is missing (#41389)

stecasta committed 38 days ago

Verified efb4cdf2

[CI] Add MTP coverage: Qwen3.5 correctness + no-sync spec decode (#40472)

stecasta committed 38 days ago

Verified 92a7c121

[DSV4] Avoid redundant dtype conversion. (#41374)

jeejeelee committed 38 days ago

Verified 307b17ce

xpu docker: pin oneAPI to 2025.3 and avoid unintended 2026 upgrade (#41380)

wendyliu235 committed 38 days ago

Verified 3ca6ca21

[CI/Build] Skip terratorch + torchgeo while PyPI has lightning quarantined (#41377)

stecasta committed 38 days ago

Verified 10558f5f

[ROCm] ROCm DeepEP API updated to latest (#39721)

itej89 committed 38 days ago

Verified 121dbe7a

[UX][Bugfix] Fix OOM by setting PyTorch `max_split_size_mb` during model loading (#41268)

MatthewBonanni committed 38 days ago

Verified f03d82ef

[EPLB] Optimize memory overhead in Nixl communicator (#40013)

ilmarkov committed 38 days ago

Verified a7fb0085

Stop mergify labelling from skipping pre-commit (#41362)

hmellor committed 39 days ago

Verified ff449b64

[Doc] Fix RTD build: pytorch.org/docs/stable/objects.inv returns 404 (#41353)

stecasta committed 39 days ago

Verified 35272295

[MoE] Make MoERunnerInterface a PluggableLayer for OOT support (#35178)

wxsIcey committed 39 days ago

Verified b55b2652

[P/D] Prefill compute optimizations with bi-directional KV cache transfers between P and D nodes (#32553)

snadampal committed 39 days ago

Verified 3179e531

[KVConnector] MultiConnector SupportsHMA (#39571)

NickLucche committed 39 days ago

Verified efdc9567

[Bugfix] correct h matrix layout in chunk_kda output kernel (#40956)

ChenxiQ committed 39 days ago

Verified 54146a9b

Fix Gemma4 MoE expert weight remapping (#41206)

Baekpica committed 39 days ago

Verified ca97f7b9

Fix Cohere ASR after HF upgrade (#40582)

ekagra-ranjan committed 39 days ago

Verified a04e0cf3

[Frontend] Add VLLM_SKIP_MODEL_NAME_VALIDATION environment variable (#34676)

dsingal0 committed 39 days ago

Verified cb1b02d0

[Bugfix] Fix persistent_topk cooperative deadlock at TopK=1024 (#41189)

zyongye committed 39 days ago

Verified a749a33d

[Refactor][kv_offload] KV Offloading maintainability improvements (#40538)

hickeyma committed 39 days ago

Verified c42981d0

[Bugfix] Fix failure to allocate KV blocks error (#41282)

wzhao18 committed 39 days ago

Verified 0ff1bf9b

[CI] Add key field to all test_areas pipeline steps (#41201)

khluu committed 39 days ago

Verified 0ab67c02

[ROCm][Bugfix][GPTOSS]: fix input_ids and expert_map args for quark w4a8 gptoss (#41165)

Rohan138 committed 39 days ago

Verified 3795d7ac

[Ci][BugFix] Fix slow DP tests due to bad teardown logic (#41166)

njhill committed 39 days ago

Verified 18599bfd

[DSv4] Use `cvt` PTX for FP32->FP4 conversion (#41015)

gau-nernst committed 39 days ago

Verified 296741d0

[Bugfix][MLA] Size arange_buffer to max_num_batched_tokens to prevent CUDA IMA (#39277)

UranusSeven committed 39 days ago

Verified a966aaed

[ROCm] Add env flags to disable dynamic MXFP4 quant and enable AITER tuned GEMMs for Attention Projection Layers (#39987)

heachary committed 39 days ago

Verified 6841f5dc

[Bugfix][Compile] Fix gc.collect/empty_cache patch arity in CUDAGraphWrapper (#41235)

roikoren755 committed 39 days ago

Verified c2fb0133

Create tests/distributed/test_mnnvl_alltoall.py (#35241)

puririshi98 committed 39 days ago

Verified ccfb620c

[BUG] Two phase pause to prevent deadlock (#39366)

hao-aaron committed 39 days ago

Verified 0335316a

Older