Commits vllm-project/vllm

type ignore

yewentao256 committed 123 days ago

b5479930

assert not None

yewentao256 committed 123 days ago

9026d5ec

update batch invariant docstring

yewentao256 committed 123 days ago

ee778e2c

[Performance] Add is_reasoning_end_streaming() override to GptOssReasoningParser (#35745)

fergusfinn committed 123 days ago

Verified 5544f8c1

[Bugfix] Fix spec decode test failures on Blackwell (SM100+) (#39546)

puririshi98 committed 123 days ago

Verified 9f39b380

[MRv2]fix: model accuracy regression caused by reusing the stale last_sampled_tokens and draft_tokens (#39833)

liuzijing2014 committed 123 days ago

Verified 9a6a66f3

Revert "[Misc] Move `pyav` and `soundfile` to common requirements" (#40276)

Isotr0py committed 123 days ago

Verified 67eb6083

Add new tp plan styles to the Transformers modelling backend (#40467)

hmellor committed 123 days ago

Verified 6ee081d1

[Model Runner V2] Multiple prompt logprobs support (#39937)

yewentao256 committed 123 days ago

Verified 66cc3fa5

Revert #38730 and #38791 (#40032)

vadiklyutiy committed 123 days ago

Verified 6d85b36a

[UX] Bump version in CG memory profiling log message (#40465)

MatthewBonanni committed 123 days ago

Verified ab5666eb

Default to 'align' mamba cache mode for Mamba-based models when speculative decoding is enabled (#40454)

roikoren755 committed 123 days ago

Verified f819265a

[MM][CG] Optimize default `max_frames_per_batch` auto-infer for ViT CUDA graph video inference (#40445)

shen-shanshan committed 123 days ago

Verified 936e0b79

[XPU][CI] Add misc, engine and lora cases on Intel GPU in CI (#39887)

zxd1997066 committed 123 days ago

Verified b2a55186

[Bugfix] LoRA: extend expert base_layer loading to Qwen3.5 and Step3.x (#37114)

HollowMan6 committed 123 days ago

Verified 908a7134

[Doc] Add Qwen3 AWQ models to documentation (#40034)

YM2132 committed 123 days ago

Verified ec5ef0ac

[Bugfix] Fix dataset name and path argument validation bug in vllm bench serve (#40288)

talorabr committed 123 days ago

Verified 7b1e0b07

Add Granite 4.1 Vision as built-in multimodal model (#40282)

artem-spector committed 123 days ago

Verified d249a9e9

[Frontend] Remove frontend pooling multi task support. (#37861)

noooop committed 123 days ago

Verified d2e2e856

feat(multimodal): support externally processed mm_kwargs with cache injection (#39502)

krishung5 committed 123 days ago

Verified 766cb65d

fix: clamp NaN/Inf in topk_softmax to prevent duplicate expert IDs (#39391)

jhaotingc committed 123 days ago

Verified 28c22215

Revert "[Startup] Parallelize torch/transformers import + weight prefetch + forkserver prewarm" (#40438)

noooop committed 123 days ago

Verified 3975eb6d

[Bugfix] Normalize malformed dict prompts that carry token IDs in `prompt` (#40339)

Alchuang22-dev committed 123 days ago

Verified 5a94a198

[Feat] dflash support for ROCm (#39703)

hangy-amd committed 123 days ago

Verified f95c11a8

[MoE] Triton MoE Perf regression - restore low latency path (#39016)

milesial committed 123 days ago

Verified 257015d5

[MM][Misc] Support image+video mixed inputs (per prompt) for VLM examples (#40335)

shen-shanshan committed 124 days ago

Verified b4784001

[Fix] Add missing space in IP fallback warning (#40359)

lesj0610 committed 124 days ago

Verified 989cc12d

[Deprecation] Deprecate cprofile and cprofile_context (#39100)

yewentao256 committed 124 days ago

Verified 301024aa

[Startup] Parallelize torch/transformers import + weight prefetch + forkserver prewarm (#40331)

simon-mo committed 124 days ago

Verified 8256833f

[Doc] Update ViT CUDA graph doc for mixed (image+video) inputs (#40355)

shen-shanshan committed 124 days ago

Verified 80975912