vllm-project/vllm

Pull Requests Commits

optimize cutlass fp8

yewentao256 committed 5 hours ago

1bfd0971

[DSv4] Drop _get_compressed_kv_buffer in DeepseekCompressor (#43690)

WoosukKwon committed 7 hours ago

Verified 193ce881

[Doc] Add line limit to AGENTS.md (#43635)

WoosukKwon committed 8 hours ago

Verified 3aea37d2

Add LM head quantization support for ModelOpt (#42124)

meenchen committed 8 hours ago

Verified 6f5b5332

[ROCm] Remove MegaMoE integration in deepseek v4 (#43629)

WoosukKwon committed 8 hours ago

Verified c8414a82

[MoE Refactor] W4a8 int8 oracle (#42789)

bnellnm committed 9 hours ago

Verified f51bbc69

[MoE Refactor] Migrate ModelOptMxFp8FusedMoE to oracle (#42768)

bnellnm committed 9 hours ago

Verified b226ddac

[Feat][DSV4] Fuse q pad into deepseek v4 fused kernel (#43162)

zyongye committed 9 hours ago

Verified 6ab6ffb4

[ROCm][CI] Extend ROCm quick reduce coverage (#40990)

AndreasKaratzas committed 10 hours ago

Verified 445ded18

[Docs][ROCm] MoRI-IO Connector Usage Guide (#43603)

simondanielsson committed 10 hours ago

Verified d565357a

Fix CuPy runtime deps and restore humming (#43530)

mmangkad committed 11 hours ago

Verified a970fb5a

[XPU] Fix fused MoE LoRA kernel crash on XPU by using platform-agnos num_compute_units (#43646)

chaojun-zhang committed 14 hours ago

Verified 861b9776

[Model] Use AutoWeightsLoader for InternLM2 (#38278)

javierdejesusda committed 14 hours ago

Verified ebd0692f

[Reasoning] [Bugfix] Reject invalid thinking_token_budget values (#43402)

linzm1007 committed 14 hours ago

Verified 739af5c7

[Misc] Support interleaved custom image benchmark datasets (#43636)

ThibaultCastells committed 14 hours ago

Verified 5d09f471

[Misc][Refactor][ROCm] Convert MoRI-related envvars to extra config args (#43303)

simondanielsson committed 14 hours ago

Verified 681d7dd3

[KV Transfer] Enable HMA by default for connectors that support it (#41847)

chfeng-cs committed 14 hours ago

Verified 755043cf

[Bugfix] Apply fc_norm in Eagle3DeepseekV2 combine_hidden_states (#43482)

yubofredwang committed 16 hours ago

Verified 97e4022c

[chores][log] change registry log from `warning` to `debug` (#43045)

ILikeIneine committed 17 hours ago

Verified b3269454

Add CuTe DSL sparse compressor support (#43584)

Jie-Fang committed 17 hours ago

Verified a37e4710

Upgrade tpu-inference to v0.20.0 (#43394)

CienetStingLin committed 18 hours ago

Verified e6adbd78

[CPU] Enable non-divisible GQA for decode workitems in mixed batches (#43032)

zhejiangxiaomai committed 18 hours ago

Verified 771e1e48

[GDN] GDN Prefill kernel for SM100 (#43273)

gau-nernst committed 18 hours ago

Verified d56612c6

[Bugfix][Model] Fix GPT2ForSequenceClassification sub-module prefix (#43579)

QingZhou-YangHY committed 19 hours ago

Verified 6f955986

[Frontend] Split the offline inference APIs and utils. (#43553)

noooop committed 19 hours ago

Verified d5cf7b4a

[Bugfix] fix device mismatch in MiniCPM-o-4_5 resampler (#43194)

yma11 committed 19 hours ago

Verified f815c999

[KV Connector] Propagate MooncakeStore load failures (#42788)

Dao007forever committed 19 hours ago

Verified c2a4005c

[KV Connector][Bugfix] MooncakeStore: don't double-apply Eagle prune in load_mask (#43516)

Dao007forever committed 19 hours ago

Verified 7966fc72

[DeepSeek V4] Move MegaMoE input prep kernel to nvidia/ops (#43632)

WoosukKwon committed 20 hours ago

Verified aa2b56ff

[LoRA] Add one shot triton kernel For MoE LoRA (#42290)

jeejeelee committed 21 hours ago

Verified ec5de7fa

Older