[SupportsQuant] Chameleon, Chatglm, Commandr (#15952)

Kyle Sayers committed over 1 year ago

[SupportsQuant] Bert, Blip, Blip2, Bloom (#15573)

Kyle Sayers committed over 1 year ago

fix: tiny fix make format.sh excutable (#16015)

yihong committed over 1 year ago

[doc] update contribution link (#15922)

Reid committed over 1 year ago

[Bugfix][V1] Fix bug from putting llm_engine.model_executor in a background process (#15367)

wwl2755 committed over 1 year ago

[misc] improve error message for "Failed to infer device type" (#15994)

youkaichao committed over 1 year ago

[Benchmark] Add AIMO Dataset to Benchmark (#15955)

Ziji Shi (Steven) committed over 1 year ago

[ROCM][V0] PA kennel selection when no sliding window provided (#15982)

Aleksandr Malyshev committed over 1 year ago

[bugfix] add seed in torchrun_example.py (#15980)

youkaichao committed over 1 year ago

fix: better error message for get_config close #13889 (#15943)

yihong committed over 1 year ago

[ROCM][KERNEL] Paged attention for V1 (#15720)

Aleksandr Malyshev committed over 1 year ago

[V1][TPU] Do not compile sampling more than needed (#15883)

Nicolò Lucchesi committed over 1 year ago

[TPU] optimize the all-reduce performance (#15903)

Chengji Yao committed over 1 year ago

[V1][TPU] TPU-optimized top-p implementation (avoids scattering). (#15736)

Hyesoo Yang committed over 1 year ago

Fix `huggingface-cli[hf-xet]` -> `huggingface-cli[hf_xet]` (#15969)

Harry Mellor committed over 1 year ago

[V1] Support Mistral3 in V1 (#15950)

Michael Goin committed over 1 year ago

[misc] instruct pytorch to use nvml-based cuda check (#15951)

youkaichao committed over 1 year ago

Restricted cmake to be less than version 4 as 4.x breaks the build of… (#15859)

Nishidha committed over 1 year ago

[Misc] V1 LoRA support CPU offload (#15843)

Jee Jee Li committed over 1 year ago

[Model] use AutoWeightsLoader in model load_weights (#15770)

rongfu.leng committed over 1 year ago