[Bug] Fix Layer `weight_block_size` Assertion Issue (#24674)

Wentao Ye committed about 1 year ago

[Doc] Clarify cudagraph capture size logic and default behavior in scheduler (#18698)

Zazzle516 committed about 1 year ago

[Qwen3-Next] MOE configs for H100 TP4 (#24699)

Chen Zhang committed about 1 year ago

[Perf] Use upstream CUTLASS for SM90 Block FP8 kernel (#23280)

Michael Goin committed about 1 year ago

[Bugfix][Attention] Fix FlashInfer MLA block size logic (#24692)

Matthew Bonanni committed about 1 year ago

[Qwen3-Next] Add B200 MoE configs for Qwen3-next (#24698)

Vadim Gimpelson committed about 1 year ago

[CI/Build] Add bc-linter to vLLM CI (#21234)

Zhewen Li committed about 1 year ago

[Qwen3-Next] MoE configs for H200 TP=1,2,4 (#24695)

Woosuk Kwon committed about 1 year ago

[Bugfix] Set `VLLM_ALLREDUCE_USE_SYMM_MEM` default to False (#24696)

Wentao Ye committed about 1 year ago

[flashinfer] [kernel] support for fp8 kv cache for trtllm prefill attention (#24197)

Xiaozhu Meng committed about 1 year ago

[Kernel][B200] `mxfp4` fused cutlass moe (#23696)

Duncan Moss committed about 1 year ago

Update Spec Decode metrics to include drafted and accepted token throughput (#24127)

Andrew Xia committed about 1 year ago

[BugFix] Fix tokenize asyncio task leak (#24677)

Nick Hill committed about 1 year ago

[Qwen3-Next] Add MoE Config for H200 (#24688)

Woosuk Kwon committed about 1 year ago

[Doc] Remove Useless Comments (#24687)

Wentao Ye committed about 1 year ago

[Ultravox] Use wrapped_model_config to instantiate inner model (#24679)

Peter Salas committed about 1 year ago

[VLM] Migrate remain DP-supported ViT models to use `disable_tp` (#24363)

Isotr0py committed about 1 year ago

[Docs] Fix formatting of transcription doc (#24676)

Harry Mellor committed about 1 year ago

fix some typos (#24616)

co63oc committed about 1 year ago

Fix model name included in responses (#24663)

Harry Mellor committed about 1 year ago