[Bugfix][Core] fix abort_seq_group and memory leak when n>1 (#14326)

courage17340 committed over 1 year ago

[Kernel] [V1] Improved performance for V1 Triton (ROCm) backend (#14152)

Thomas Parnell committed over 1 year ago

[Doc] Correct beam_search using in generative_models.md (#14363)

Irina Yuryeva committed over 1 year ago

[Doc] Update reasoning with stream example to use OpenAI library (#14077)

Yanyi Liu committed over 1 year ago

[Frontend][Docs] Transcription API streaming (#13301)

Nicolò Lucchesi committed over 1 year ago

[Core] Optimizing cross-attention `QKVParallelLinear` computation (#12325)

Nicolò Lucchesi committed over 1 year ago

[V1][VLM][Pixtral-HF] Support Pixtral-HF on V1 (#14275)

lkchen committed over 1 year ago

[Model] Update Paligemma multimodal processing with PromptUpdate (#14015)

kYLe committed over 1 year ago

[Hardware] Update the flash attn tag to support Blackwell (#14244)

Pavani Majety committed over 1 year ago

[Bugfix][CI] ALiBi test case in xformers multi_query_kv_attention (#11301)

Nicolò Lucchesi committed over 1 year ago

[V1] LoRA - Enable more V1 tests (#14315)

Varun Sundar Rabindranath committed over 1 year ago

[Bugfix][Structured Output] Support outlines engine with reasoning outputs for DeepSeek R1 (#14114)

Ce Gao committed over 1 year ago

[misc] Mention `ray list nodes` command to troubleshoot ray issues (#14318)

Rui Qiao committed over 1 year ago

[BugFix] MLA + V1, illegal memory access and accuracy issues (#14253)

Lucas Wilkinson committed over 1 year ago

[Build] Add UV_HTTP_TIMEOUT to avoid timeout during installation (#13850)

Yuan Tang committed over 1 year ago

Add benchmark for DeepGEMM and vLLM Block FP8 Dense GEMM (#13917)

Michael Goin committed over 1 year ago

[CI/Build] Use spawn multiprocessing mode for V1 test pipeline (#14243)

Russell Bryant committed over 1 year ago

[BugFix] Fix prefix caching V0 MLA (#14255)

Lucas Wilkinson committed over 1 year ago

[Bugfix] Remove num_tokens_across_dp (#14302)

Tyler Michael Smith committed over 1 year ago

[Bugfix] Fix DeepSeek MTP crash when using TP1ModelRunner with CUDA graph due to shape mismatch (#14237)

pyc96 committed over 1 year ago