[feat] Add FP8 per-tensor Q scale support to Triton attention backend (#42080)

Dom Brown committed 6 months ago

[Misc] add humming to dependencies (#42540)

Jinzhen Lin committed 6 months ago

[Refactor] Extract extract_types_from_schema utility from Minimax M2 tool parser (#43025)

Flora Feng committed 6 months ago

[Model] Openvla support (#42654)

Wang Yiwen committed 6 months ago

[XPU] update xpu graph usage (#43043)

Xinyu Chen committed 6 months ago

[Perf][4/n] Eliminate various GPU<->CPU syncs (#42347)

Nick Hill committed 6 months ago

[Bugfix] Fix top logprobs token placeholders in `/inference/v1/generate` (#42887)

Sage committed 6 months ago

[CI] Add MTP + PD disagg test for Qwen3.5 (#42677)

zhanqiuhu committed 6 months ago

[Misc][MM] Remove redundant code in CLIPAttention (#43046)

Shanshan Shen committed 6 months ago

[Model Refactoring] Rename deepseek_v4.py to model.py [4/N] (#43077)

Woosuk Kwon committed 6 months ago

[KVConnector][DSV4] HMA support for Mooncake store connector (#42828)

Yifan Qiao committed 6 months ago

[bug] AsyncScheduler drops first post-resume token after pause_generation + clear_cache (#42117)

Aaron Hao committed 6 months ago

[Model Refactoring] Move deepseek_v4_ops to models/deepseek_v4 [3/N] (#43073)

Woosuk Kwon committed 6 months ago

[Frontend] Consolidate beam search by BeamSearchMixin. (#42946)

wang.yuqi committed 6 months ago

[Docs] Reorganize online serving docs. (#41907)

wang.yuqi committed 6 months ago

[Misc] Aligning tokwise pooler heads for consistency (#43041)

Taneem Ibrahim committed 6 months ago

[XPU] Use custom op collective behavior (#41354)

Chaojun Zhang committed 6 months ago

[Docs] Add SVG images for pooling models. (#42626)

Gracie Guo (UX) committed 6 months ago

fix: use keyword arguments for shard_id and expert_id in weight_loade… (#42671)

Junyan Xu committed 6 months ago

[ci] Route 28 gpu_1_queue tests to h200_35gb queue (#43030)

Kevin H. Luu committed 6 months ago