[Doc] Add documentation for vLLM continuous benchmarking and profiling (#25819)

Naman Lalit committed about 1 year ago

[BugFix][torch.compile] KV scale calculation issues with FP8 quantization (#25513)

Adrian Abeyta committed about 1 year ago

[Bugfix] Use correct key "ignore" for config.json non-quantized layers (#25706)

Lee Nau committed about 1 year ago

[Model] Remove MotifForCausalLM (#25866)

Jee Jee Li committed about 1 year ago

[CI/Build] Include Transformers backend test in nightly transformers test (#25885)

Isotr0py committed about 1 year ago

[Bugfix][Speculative Decoding] Fix Eagle3 quantization config issue (#25883)

Rahul Tuli committed about 1 year ago

[Nixl][P/D] Add cuda2cpu support (HD->DH transfer) (#24690)

Chenxi Yang committed about 1 year ago

[torch.compile] serialize cudagraph_mode as its enum name instead of value (#25868)

Jiangyun Zhu committed about 1 year ago

[Model][Bugfix] Fix issues in MiDashengLM implementation for quantized models (#25854)

Zhou Jiahao committed about 1 year ago

[Bugfix] Fix requirements paths in install instructions (#25827)

Yingjun Mou committed about 1 year ago

update to latest deepgemm for dsv3.2 (#25871)

youkaichao committed about 1 year ago

[Misc] Remove more `get_input_embeddings_v0` (#25857)

Cyrus Leung committed about 1 year ago

[V0 Deprecation][Models] Remove all V0 condition for mm embeddings merge (#25331)

Isotr0py committed about 1 year ago

[Bugfix] Fallback ViT attn backend to SDPA for blackwell (#25851)

Roger Wang committed about 1 year ago

[XPU]Fix xpu spec decoding UTs, avoid using cuda graph (#25847)

Kunshang Ji committed about 1 year ago

Add Phi4FlashForCausalLM to _PREVIOUSLY_SUPPORTED_MODELS (#25832)

Thomas Parnell committed about 1 year ago

[P/D] NIXL Updates (#25844)

Robert Shaw committed about 1 year ago

[Misc] fix tests failure by using current_platform (#25825)

Juechen Liu committed about 1 year ago

Remove redundant cudagraph dispatcher warning (#25841)

Michael Goin committed about 1 year ago

[Bugfix] fix Qwen3VLMoe load when pp > 1 (#25838)

JJJYmmm committed about 1 year ago