A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 6, 19:15 UTC
main
[Perf] Add TRTLLM FP8 MoE Modular Kernel (#36307)
Wei Zhao committed 8 months ago
[CI] Fix mypy for vllm/reasoning (#35742)
Martin Hickey committed 8 months ago
[docs] Add lightweight AI assisted contribution policy (#30947)
Mark McLoughlin committed 8 months ago
[Perf] add packed recurrent fast path for decode (#36596)
caozuoba committed 8 months ago
[Frontend] Split `OpenAIServingModels` into `OpenAIModelRegistry` + `OpenAIServingModels` (#36536)
Sage committed 8 months ago
[Bugfix] Fix crash when tool_choice=required exceeds max_tokens (#36841)
Chauncey committed 8 months ago
[MM][OOT] Support CPU `seq_lens` for OOT MMEncoderAttention kernels (#36605)
Shanshan Shen committed 8 months ago
[UX] Only show FP4 Marlin fallback warning for w4a4 models (#36806)
Michael Goin committed 8 months ago
[Bugfix] Warm up Triton autotuner for GDN layers during V1 profiling (#36599)
Xu Jinyang committed 8 months ago
[KV Connector] Support using FlexKV as KV Cache Offloading option. (#34328)
sfeiqiang committed 8 months ago
[Bugfix][Model] Fix DeepSeek-OCR TensorSchema crash on empty images_crop (#36670)
István Ketykó committed 8 months ago
[Frontend] OpenAI Responses API supports Tool/Function calling with streaming (#29947)
Chauncey committed 8 months ago
[Tests] Skip model weight download for render-only test server (#36813)
Sage committed 8 months ago
replace `with torch.cuda.device` with `with torch.accelerator.device_index` (#36144)
Yan Ma committed 8 months ago
[Kernel][Helion][13/N] Force static_shapes=False in helion register (#36677)
Yanan Cao committed 8 months ago
[BugFix] Fix multiple/duplicate stdout prefixes (#36822)
Nick Hill committed 8 months ago
[CI Failure] Fix Language Models Test (Extended Pooling) daily CI Failure (#36761)
wang.yuqi committed 8 months ago
[Model Runner V2] Do not initialize sampler for non-last PP ranks (#36824)
Woosuk Kwon committed 8 months ago
[LMCache] Fault Tolerance Mechanism (#36586)
Yuwei An committed 8 months ago
more models for vLLM Benchmark Suite (#35086)
Louie Tsai committed 8 months ago