vllm-project/vllm

Pull Requests Commits

low latency combine

zhuohan123 committed 49 days ago

99e2379b

low latency dispatch

zhuohan123 committed 49 days ago

da26dce7

refactor and add low latency code

zhuohan123 committed 49 days ago

48dcc72d

add combine example

zhuohan123 committed 49 days ago

e3e2bb38

zhuohan123 committed 50 days ago

4e2abe99

complicated assert for correctness check

zhuohan123 committed 50 days ago

177f5d75

deepep HT dispatch no abstraction

zhuohan123 committed 52 days ago

dcf059ab

zhuohan123 committed 54 days ago

98e71a49

batched_deepgemm_contiguous

zhuohan123 committed 54 days ago

1f4472ba

add triton_group_gemm_masked

zhuohan123 committed 54 days ago

850876a1

Add contiguous triton moe example

zhuohan123 committed 54 days ago

a608dfab

[DO NOT MERGE] Experiments related to MoE kernels

zhuohan123 committed 55 days ago

2d4215b9

[CI/Build] Add tool to build vllm-tpu wheel (#19165)

mgoin committed 57 days ago

Verified 7ef60528

[easy] fix pre commit error on trunk (#26665)

hl475 committed 57 days ago

Verified 4fca1a1b

[Models][Qwen3VL] Speedup `fast_pos_embed_interpolate` (#26647)

lgeiger committed 57 days ago

Verified a6049be7

[Feature] Add support for naver/splade-v3 (BERT-based sparse embedding model) (#26339)

gjgjos committed 57 days ago

Verified 18ed7746

Update `Optional[x]` -> `x | None` and `Union[x, y]` to `x | y` (#26633)

hmellor committed 57 days ago

Verified 8fcaaf6a

[Bugfix] Fix GPU_ID issue in test script (#26442)

xuechendi committed 57 days ago

Verified 9bb38130

[Bugfix][DCP] Set default CUDAGraphMode to PIECEWISE for DCP (#26574)

FENP committed 57 days ago

Verified b91d8db8

[Bugfix][CI/Build] Fix failing Mteb CI (#26638)

Isotr0py committed 57 days ago

Verified 045b396d

[MISC] Rename the torch profiler filename as instance_id+rank_id for merging the Profiler results of each Rank (#25867)

noooop committed 57 days ago

Verified 76852017

[PERF] [Qwen3-next] Speed up gated RMSNorm (#26207)

vadiklyutiy committed 57 days ago

Verified 82e64c7a

Add @noooop to codeowner for pooling models (#26652)

noooop committed 58 days ago

Verified 4ca20405

[EPLB] Support ernie4.5-moe (#22100)

HsChen-sys committed 58 days ago

Verified c5c8f5ea

[compile] Fix inductor partition config (#26645)

angelayi committed 58 days ago

Verified 01653a91

CP: make correct_attn_out robust to 4‑D views and fix Triton arg binding (#26509)

hl475 committed 58 days ago

Verified 0cd103e7

[Benchmark] Support Infinity API (#26641)

DarkLight1337 committed 58 days ago

Verified 5be7ca1b

[Bugfix] Fix qwen-moe packed_modules_mapping (#26634)

jeejeelee committed 58 days ago

Verified f0a30a06

[Bugfix][Qwen3VL] fix deepstack in qwen3vl (#26626)

JJJYmmm committed 58 days ago

Verified 9d6cff3e

[compile] Add patched_fused_scaled_matmul_reduce_scatter (#26604)

angelayi committed 58 days ago

Verified a25f2ade

Older