vllm-project/vllm

Pull Requests Commits

remove dead code util

yewentao256 committed 1 day ago

a1bf5c80

[ROCm][K3] Extend FP8 asm MLA prefill to non-divisor small head counts (#51040)

xiaohuguo2023 committed 1 day ago

Verified 1a085dad

[Bugfix][DP] Synchronize the device on pause completion (#52914)

aoshen02 committed 1 day ago

Verified 17da4859

[Bugfix][ROCm][Disagg] Fix MoRIIO shared KV memory region registration (#53698)

avininjamay8 committed 1 day ago

Verified 8d301f07

[Bugfix] Thread kv_transfer_params into engine for /inference/v1/generate (disagg) (#42644)

hallerite committed 1 day ago

Verified c71f6f8a

[Bugfix] Make Gemma4 MTP suppress_tokens masking CUDA-graph-safe (#53884)

lucianommartins committed 1 day ago

Verified 2ab18743

[ROCm][DSV4][Perf] Fuse DeepSeek V4 C4 compressor GEMMs (#53838)

Fangzhou-Ai committed 1 day ago

Verified 657f9b9c

[Bugfix][ROCm] Capture CUDA graphs on the current stream (#53818)

fanxingran committed 1 day ago

Verified 080a66a6

[Model] Pixtral: use packed multimodal encoder attention (#52185)

oliverholworthy committed 1 day ago

Verified 28b484e5

[Mypy Fix] Mypy fix for "vllm/model_executor/models/[tT]" (#53466)

ZHIHANCHEN03 committed 1 day ago

Verified c47ca4aa

[Bugfix] Handle empty FlatLogprobs slices and delta output (#53704)

new-TonyWang committed 1 day ago

Verified 903a0219

[DeepEPv2] Support MXFp8 Activation Scale Dispatch (#51398)

robertgshaw2-redhat committed 1 day ago

Verified 6a5e8f59

[Doc] Fix local input path in run-batch examples across docs (#53220)

qwerqwerqwe8688-jpg committed 1 day ago

Verified 15baeaee

[MoE] enable all2all fi_one_sided by default (#53311)

arpera committed 1 day ago

Verified 2cd6c664

[CI/Build] Improve pre-commit fail message (#53866)

DarkLight1337 committed 1 day ago

Verified 31739ceb

[Offloader] Offload submodules that make_layers never reaches (#53120)

ray24777 committed 1 day ago

Verified 61d4f566

[CI] forward fix CRCR report step in the torch-nightly lane (#53732)

atalman committed 2 days ago

Verified 7a999387

[CI] Build mamba-ssm with C++20 for torch 2.14 nightly compatibility (#49600)

atalman committed 2 days ago

Verified b821d7b2

[Bugfix][Processor] Replace bare asserts with ValueError in DeepseekVLV2/OCR processors (#53854)

adisivaprasad committed 2 days ago

Verified 73bd7c83

[AttentionBackend][HPC-ops] update hpc rope norm to support stride kv cache (#53705)

thisjiang committed 2 days ago

Verified 2267d3b1

[LoRA] Cleanup VocabParallelEmbedding (#53843)

jeejeelee committed 2 days ago

Verified 75aa189b

[Config] Delegate PCP compatibility checks to PCP manager (#53853)

pisceskkk committed 2 days ago

Verified e376d45e

[Bugfix][Mooncake] Fix Mamba prefill truncation ordering (#53663)

ZeldaHuang committed 2 days ago

Verified f14369c7

[CI/Build][The Rock] Use model_class_overrides so spawned worker can use test PredictableLlamaForCausalLM class when worker spawned using Python 3.14 (#49218)

rasmith committed 2 days ago

Verified cde7ba92

[Hardware][AMD][Perf][Bugfix] Update ROCr and clr in base image (#53712)

mawong-amd committed 2 days ago

Verified 796822d1

[RL] Add rank-local IPC weight updates (#52497)

aoshen02 committed 2 days ago

Verified ee5ae2ae

[Bugfix][Model] Honor Molmo2 dummy video num_frames >= 2 override (#53830)

hungnnvidia committed 2 days ago

Verified a5f82882

BugFix(PCP): use persistent input buffers for PIECEWISE CUDA graphs (#53515)

pisceskkk committed 2 days ago

Verified b1fbbc2a

[Bugfix][Structured Output] Preserve reasoning adapters for shared parser engines (#52830)

sfeng33 committed 2 days ago

Verified 46638857

[Kernel][Perf] Tune fused_moe FP8 config for Qwen3.5 on L40S (+7%) (#53819)

nicholaskh-ai committed 2 days ago

Verified 044b0522

Older