[EP] Add logging for experts map (#22685)

22quinn committed about 1 year ago

[Fix] remove is_marlin param in benchmark_moe (#23286)

shixianc committed about 1 year ago

Feature/mla tests (#23195)

Matthew Bonanni committed about 1 year ago

Remove chunked_prefill_enabled flag in V1 MLA (#23183)

Matthew Bonanni committed about 1 year ago

[misc] fix multiple arch wheels for the nightly index (#23110)

youkaichao committed about 1 year ago

[Feature] use --eplb_config to set eplb param (#20562)

rongfu.leng committed about 1 year ago

Do not use eval() to convert unknown types (#23266)

Russell Bryant committed about 1 year ago

[Perf] Speed up function `_convert_tokens_to_string_with_added_encoders` by 13.7x (#20413)

Saurabh Misra committed about 1 year ago

[Frontend] improve error logging of chat completion (#22957)

Chen Zhang committed about 1 year ago

[Fix] correct tool_id for kimi-k2 when use tool_choice=required (#21259)

bigmoyan committed about 1 year ago

[Kernel/Quant] Remove the original marlin format and qqq (#23204)

Michael Goin committed about 1 year ago

Small fix for Command-A-Vision (#23268)

dongluw committed about 1 year ago

Limit HTTP header count and size (#23267)

Russell Bryant committed about 1 year ago

Migrate Mistral3ImagePixelInputs to TensorSchema (#21945)

Benji Beck committed about 1 year ago

[torch.compile] Support conditional torch.compile per module (#22269)

Yong Hoon Shin committed about 1 year ago

[FIXBUG ] Allow disabling rocm_aiter_fa backend for ROCm GPUs not compatible with AITER (#22795)

JartX committed about 1 year ago

[Misc] Add max_seq_len to CommonAttentionMetadata (#23216)

Woosuk Kwon committed about 1 year ago

[CLI][Doc] Formalize `--mm-encoder-tp-mode` (#23190)

Cyrus Leung committed about 1 year ago

[Kernel] CUTLASS MoE FP8: Integrate cuda moe permute/unpermute (#23045)

shixianc committed about 1 year ago

[Bugfix] Ensure correctness of HCXVision processing (#23254)

Cyrus Leung committed about 1 year ago