vllm-project/vllm

Pull Requests Commits

[Build] Test DeepGEMM FP8 fix on nightly baseline

khluu committed 5 days ago

1e9a93ae

[XPU][CI]Adjust Samplers test ENV for Intel GPU (#50199)

zxd1997066 committed 5 days ago

Verified cb810483

[Bugfix][Reasoning] kimi_k3: O(delta) reasoning-end check on the decode path (#50886)

abmfy committed 5 days ago

Verified adbf08d9

[ROCm][AITER] Hotfix for `memory access fault` errors in AITER triton MOE routing (#50859)

fxmarty-amd committed 5 days ago

Verified 385d4c08

[CI][Bugfix] Fix flaky `test_store_orders_after_compute_write` (#50926)

njhill committed 5 days ago

Verified 74295e3b

Perf/h20 moe config e256 n512 (#48825)

zzt93 committed 5 days ago

Verified 413e70d5

[ROCm][Test] Use BF16 for Jina v5 nano MTEB test (#50917)

AndreasKaratzas committed 5 days ago

Verified 8adc840c

[Kimi-K3] Migrate FlashKDA to PyTorch stable ABI (#50818)

gau-nernst committed 5 days ago

Verified 11f88260

[Bugfix][Kimi-K3] Enforce packed rows and op availability in AttnRes dispatch (#50567)

namgyu-youn committed 5 days ago

Verified 41ba11b8

llmd+vllm+mori-ep(inter node wide-ep)+mori-io(write) for 2p2d with dp=ep=16 tp=1 (#45043)

shikamd123 committed 5 days ago

Verified f4276120

[Kernel] Extend CuTe DSL skinny GEMM to GLM-5.2 (#49791)

zhou9402 committed 5 days ago

Verified c8109375

[Bugfix] Fix Qwen3-Omni crash on video with no audio track when use_audio_in_video=True (#48420)

RyanJHamby committed 5 days ago

Verified 6a9109d8

[Bugfix][Hybrid] Fix cross-block race on num_accepted in MRv2 align prefix cache (#50432)

fuscof-ibm committed 5 days ago

Verified c2881ce6

fix: NVFP4 quantization out_dtype should match model dtype, not torch default (#48861)

fattchris committed 5 days ago

Verified 0b37d838

feat(frontend): session id plumbing into requests (#48048)

karen-sy committed 5 days ago

Verified f57123aa

[ModelRunnerV2] Fix scalar Mamba state update with int32 mappings (#50327)

shenoyvvarun committed 5 days ago

Verified e578de31

[ROCm] Enable AITER and FP8 inference on GFX120x (#43615)

skysnow2001 committed 5 days ago

Verified f43e1d26

[Bugfix][Model Runner V2] Restore multimodal draft capability detection (#50417)

TQCB committed 5 days ago

Verified 4a3447d2

[MRV2] Enable routed-experts capture (#50721)

aoshen02 committed 5 days ago

Verified 42ab184e

[ROCm][Test] Fix AITER MXFP4 oracle contract (#50728)

AndreasKaratzas committed 5 days ago

Verified e279f715

[Bugfix][Frontend] Reject empty gRPC stop strings (#50746)

zcxGGmu committed 5 days ago

Verified c4e9f09d

[Hybrid] Stage the postprocess inputs with a single loop over the request list (#48120)

fuscof-ibm committed 5 days ago

Verified 755513d9

[Bugfix] Default Gemma3 Model intermediate_tensors to None (#50777)

taneem-ibrahim committed 5 days ago

Verified 1c0d2079

[Bugfix] Emit a valid media type from encode_{audio,image,video}_url (#49056)

vineethsaivs committed 5 days ago

Verified 7f2e78ba

[ROCm][Kimi-K3] aiter moe environment variable cleanup (#50582)

hongxiayang committed 5 days ago

Verified 8f50685c

[Bugfix] Validate NIXL speculative config compatibility (#49230)

tzulingk committed 5 days ago

Verified 952694e3

[Kernel] Skip fully masked key blocks in windowed Triton prefill (#50776)

almogtavor committed 5 days ago

Verified 65cf1276

[CI][ROCm] Export Helion benchmark script in test artifacts (#50726)

AndreasKaratzas committed 5 days ago

Verified 76d995df

[Bugfix] Shard UniformTypeKVCacheSpecs block table width under DCP (#50823)

drakosha committed 5 days ago

Verified f0de1a60

[Kimi-K3] Add option to shard the shared expert instead of replicating (#50656)

tlrmchlsmth committed 5 days ago

Verified 5df9999f

Older