vllm-project/vllm

Pull Requests Commits

reuse cuda major

NickLucche committed 11 days ago

d423438f

NickLucche committed 12 days ago

31a31036

[Bugfix] use `served_model_name` for multimodal error message (#41003)

msanft committed 12 days ago

Verified 2c06cf34

Deprecate support for Transformers v4 (#40389)

hmellor committed 12 days ago

Verified e6f710a8

[Model] Add MiMo-V2.5 support (#40967)

Isotr0py committed 12 days ago

Verified c245d35f

[Bugfix] Fix k_norm weight sharding in MiniMaxM2Attention when total_num_kv_heads < tp_size (#38191)

wxsIcey committed 12 days ago

Verified f8ac0c7c

Add system_fingerprint field to OpenAI-compatible API responses (#40537)

simon-mo committed 13 days ago

Verified ebf862c3

[Examples] Resettle generate examples. (#36464)

noooop committed 13 days ago

Verified 8d8062d0

[Bugfix] Install libcublas-dev in Dockerfile for FlashInfer CuTe DSL JIT (#39855)

esmeetu committed 13 days ago

Verified 98596134

[DSV4] Add silu clamp limit to shared expert (#40950)

zyongye committed 13 days ago

Verified 706a04d3

[Bugfix] Remove invalid deepstack boundary check for Qwen3-VL (#40932)

Isotr0py committed 13 days ago

Verified 22631f80

[Attention][TurboQuant] Share dequant buffers, eliminate float16_copy (#40941)

bhoomit committed 13 days ago

Verified 2cc008e7

[Perf] FP8 FlashInfer Attn for ViT (#38065)

zhandaz committed 13 days ago

Verified 5d5c7764

Cutlass W4A16 (Machete) Tests (#35450)

ojhaanshika committed 13 days ago

Verified 592ae680

[Bugfix] Cap SWA/chunked-local runtime admission to startup pool-sizing bound (#40946)

Dao007forever committed 13 days ago

Verified 7b1bc0a3

[Tests] Gate Isaac under Transformers v5 (#40907)

SiluPanda committed 13 days ago

Verified c0879d94

[Model Runner V2] Fix rejection sampling acceptance rate gap vs MRV1 (#40651)

TheEpicDolphin committed 13 days ago

Verified f5f98785

Auto-disable expandable_segments around cumem memory pool (#40812)

youkaichao committed 13 days ago

Verified 2ce95a76

[Feat] DeepSeek V4 Rebased (#40860)

ivanium committed 13 days ago

Verified 4d51588e

[torch.compile]: Disable Sequence Parallelism (SP) for piecewise compilation (#38373)

SouthWest7 committed 13 days ago

Verified 32e45636

[KV Offload] Offload all KV blocks when doing prefill in P/D (#40346)

omerpaz95 committed 14 days ago

Verified b39c266d

[Bugfix] Size FlashInfer NVLink MNNVL workspace to EP group (#40893)

Dao007forever committed 14 days ago

Verified 9558f439

[LoRA] MoE LoRA Refactor (#40338)

jeejeelee committed 14 days ago

Verified 8cd174fa

[Bugfix] Fix the DSML token leakage in DSV4/3.2 (#40806)

chaunceyjiang committed 14 days ago

Verified c798593f

[Bugfix][MoE] Only unpad routed output before shared expert add or routed output transform (#40865)

netanel-haber committed 14 days ago

Verified 12a3f645

[kv_offload+HMA][11/N]: Support store with multiple KV groups (#39403)

orozery committed 14 days ago

Verified 60cd878a

[CI][AMD]BugFix] Fix deadlock occuring in test_moe_layer (#40767)

rasmith committed 14 days ago

Verified 1e9f19ca

[Opt] Optimize deepstack buffer handling for multimodal Qwen3 models (#40145)

labAxiaoming committed 14 days ago

Verified 6646c0c7

[ROCm][Engine] Fix GPU memory leaks in engine shutdown and test workaround for async KV prefix cache reset (#38503)

AndreasKaratzas committed 15 days ago

Verified 95995bbe

[Feature] Warm up readonly multimodal processor during renderer startup (#40797)

fake0fan committed 15 days ago

Verified 07351e08

Older