Commits vllm-project/vllm

Merge branch 'main' into zhuohan/remove-unnecessary-instance_id-setup

DarkLight1337 committed 24 days ago

Verified 423ff4eb

[Bugfix] Warm up Triton autotuner for GDN layers during V1 profiling (#36599)

AuYang261 committed 24 days ago

Verified 3e64fe4a

[KV Connector] Support using FlexKV as KV Cache Offloading option. (#34328)

feiqiangs committed 24 days ago

Verified 8cb24d3a

[Bugfix][Model] Fix DeepSeek-OCR TensorSchema crash on empty images_crop (#36670)

ketyi committed 24 days ago

Verified 00726c74

[Frontend] OpenAI Responses API supports Tool/Function calling with streaming (#29947)

chaunceyjiang committed 24 days ago

Verified 9fe404ed

[Tests] Skip model weight download for render-only test server (#36813)

sagearc committed 24 days ago

Verified 802f306c

replace `with torch.cuda.device` with `with torch.accelerator.device_index` (#36144)

yma11 committed 24 days ago

Verified 894843eb

[Kernel][Helion][13/N] Force static_shapes=False in helion register (#36677)

gmagogsfm committed 24 days ago

Verified 584a3f56

[BugFix] Fix multiple/duplicate stdout prefixes (#36822)

njhill committed 24 days ago

Verified 36735fd7

[CI Failure] Fix Language Models Test (Extended Pooling) daily CI Failure (#36761)

noooop committed 24 days ago

Verified 6ecabe49

[Model Runner V2] Do not initialize sampler for non-last PP ranks (#36824)

WoosukKwon committed 24 days ago

Verified 2f8b4ce0

[LMCache] Fault Tolerance Mechanism (#36586)

Oasis-Git committed 24 days ago

Verified 2ef69456

more models for vLLM Benchmark Suite (#35086)

louie-tsai committed 24 days ago

Verified 17852aa5

[Bugfix] Fix minimax_m2 tool parser when stream interval > 1 (#35895)

sfeng33 committed 24 days ago

Verified 8647c6cf

[XPU][Doc] Remove manual OneAPI install step, now handled by torch-xpu (#36831)

jikunshang committed 24 days ago

Verified 513949f9

[Frontend] Exclude anthropic billing header to avoid prefix cache miss (#36829)

njhill committed 24 days ago

Verified 262b76a0

[Perf] Optimize compute maxsim using batched version, 3.2% E2E throughput improvement (#36710)

yewentao256 committed 24 days ago

Verified c34ba6b9

[ROCm][CI/Build] Add gfx1152/gfx1153 (Krackan) to HIP supported architectures (#36499)

mgehre-amd committed 24 days ago

Verified 24062b70

[BUG] Fix async rlhf tests (#35811)

hao-aaron committed 24 days ago

Verified d6b61e51

[Kernel] [Helion] [15/N] Split config files into per-platform files (#36698)

gmagogsfm committed 24 days ago

Verified cf632499

[Kernel] [Helion] [12/N] Use FakeTensorMode to avoid GPU allocation during config key computation (#36563)

gmagogsfm committed 24 days ago

Verified a3774a81

[Kernel] [Helion] [14/N] Set autotune_ignore_errors=True during autotuning (#36683)

gmagogsfm committed 24 days ago

Verified 0ce21c46

[Model Runner V2] Add WhisperModelState [6/N] (#35790)

WoosukKwon committed 24 days ago

Verified 55eed6b7

[Model Runner V2] Add probabilistic rejection sampling for spec decoding (#35461)

TheEpicDolphin committed 24 days ago

Verified c77181e5

[LMCache] Pass TP size in lookup for MLA multi-reader locking (#36129)

maobaolong committed 24 days ago

Verified 12001f2e

[BugFix][kv_offload] Fix offloading decodes with async scheduling (#33881)

orozery committed 24 days ago

Verified 7ee5d509

[Bugfix][ROCm] Strip block_size before attention backend validation (#36274)

jennyyyyzhen committed 24 days ago

Verified 428bc718

[BugFix]: add bagel to MM_PREFIX_LM_MODELS (#36316)

princepride committed 24 days ago

Verified ff1e3d9c

[Refactor] Remove dead code in KV connector (#36424)

yewentao256 committed 24 days ago

Verified 35bdca54

[ROCm] add tuned moe_wna16_triton kernel configs for CDNA4 (#35093)

amd-asalykov committed 24 days ago

Verified 8a248427