Commits vllm-project/vllm

[Bugfix] Bound recursive argument parsers

BugenZhao committed 2 hours ago

Verified efc92458

[Bugfix][Distributed] Gate cross-node MNNVL custom all-reduce by group capability (#53253)

JulianZJN committed 3 hours ago

Verified 99013d77

[Bugfix] Fix ncclCommQueryProperties heap overflow with NCCL >= 2.31 (#53008)

Xuan-1998 committed 3 hours ago

Verified 026d5af7

fix: improve token_ids_cpu swap to copy only valid indices (#36255)

zzaebok committed 3 hours ago

Verified 68c52b5a

[Rust Frontend] Add support for `truncate_prompt_tokens` and `truncation_side` (#48584)

pranavthakur0-0 committed 3 hours ago

Verified 085e9bb0

[CI][Test] Deflake the rms_norm scaling-property assertions (#54271)

okorzh-amd committed 3 hours ago

Verified 1a16c2ad

[Bugfix] Set breakable graph env before Ray actor import (#53293)

alexeldeib committed 4 hours ago

Verified 9662ab08

[ROCm][PERF] Enable AITER PA gluon decode for MiniMax-M3 MTP and dense layers (#52849)

ukannika committed 5 hours ago

Verified ae5b8e4a

[CI/Build] Add advisory PR title format check (#54263)

BugenZhao committed 5 hours ago

Verified b131311f

[Model] Support speculative decoding method for PLaMo3 (#54239)

Alnusjaponica committed 5 hours ago

Verified df14152a

[ROCm][CI] Fix test_ray_v2_executor (#54249)

charlifu committed 5 hours ago

Verified fd98d32f

[Bugfix] Keep the Moondream3 MoE all-reduce out of the fused-path try (#54152)

okorzh-amd committed 5 hours ago

Verified c274d361

[Core][KV Connector] Start async KV loads after the forward launch when no sync loads are scheduled (#53333)

GirasoleY committed 5 hours ago

Verified 2aac565c

[Bugfix][MRV2] Release layer-bound KV cache memory in shutdown() (#54246)

njhill committed 6 hours ago

Verified 96242aa5

[Bugfix][ROCm] Pre-allocate `wvSplitKrc` static workspaces before KV init (#54247)

njhill committed 6 hours ago

Verified 70732942

[Mypy] Fix mypy typing for model interfaces and H/I models (#54079)

taneem-ibrahim committed 7 hours ago

Verified 21fa2c5a

[Rust Frontend] Optimize SSE streaming hot path (#51321)

reidliu41 committed 7 hours ago

Verified 74850f9f

[Bugfix] Fix int32 token offset overflow in fused SiLU block quant (#53409)

canlahlah committed 7 hours ago

Verified 06cccf87

[BugFix] Bind RayExecutorV2 TCPStore before publishing its port (#50969)

aoshen02 committed 8 hours ago

Verified 67e86d1e

[ROCm] remove VLLM_ROCM_USE_AITER_FP4_ASM_GEMM environment variable; make w4a4 use the preshuffle triton+asm by default (#53141)

afriedri committed 8 hours ago

Verified 2f1cba79

update quark docs to include online quantization (#52736)

hangy-amd committed 9 hours ago

Verified 31c57950

[Kimi-K3][Kernel] Optimize the low-M fused latent MoE tail (#54168)

zyongye committed 10 hours ago

Verified d9dabfa3

[Perf] Split xdrope_positions H2D copy into per-row transfers (#53412)

JasonKeyiL committed 10 hours ago

Verified 6f91e3d9

[MM][CG] Support ViT full CUDA graph for Idefics3 and SmolVLM (#47625)

CHIPMUNK-T0T committed 10 hours ago

Verified ffe690ec

[ROCm][Quantization][MOE] Enable fused shared experts for block-quantized FP8 (#53097)

xuebwang-amd committed 11 hours ago

Verified 06569a86

[Rust Frontend] Reduce copy in auxiliary frame resolution (#54148)

BugenZhao committed 11 hours ago

Verified 11012d2a

[XPU]bump up vllm_xpu_kernels to 0.1.14.1 (#54203)

xinyu-intel committed 13 hours ago

Verified 94a54f58

[Bugfix] Restore multimodal support on the plain "vllm" throughput backend (#52168)

mganczarenko committed 17 hours ago

Verified d1922cb5

[ROCm][CI] Warm up the RLHF dev server before the pause/resume timing checks (#53594)

stefankoncarevic committed 18 hours ago

Verified 5f213ed1

[Bugfix][CI/Build] Fail closed when selected precompiled CUDA variant is unavailable (#52545)

jungjiyu committed 18 hours ago

Verified 3bb19cd8