[All Reduce] Change default backend of Flashinfer All Reduce to trtllm (#35793)

Hanjie Qiu committed 8 months ago

[Bugfix] Fix MM processor test for Qwen3.5 (#35797)

Roger Wang committed 8 months ago

[MoE][Perf] Wrap DSV3 QKVAProj GEMM in custom op for torch.compile (#35751)

Robert Shaw committed 8 months ago

clean unused cudagraph_batch_sizes (#35552)

Boyuan Feng committed 8 months ago

[Bugfix] Fix missing sequence_lengths in qwen3_omni_moe_thinker (#35741)

Ye (Charlotte) Qi committed 8 months ago

[BUG] Fix rlhf_async example (#35788)

Aaron Hao committed 8 months ago

[ci] Add Ray compatibility check informational CI job (#34672)

Jeffrey Wang committed 8 months ago

[KVConnector] Auto-downgrade to PIECEWISE cudagraph mode for layerwise async ops (#31057)

Yashwant Bezawada committed 8 months ago

[Spec Decode] Add hidden states extraction system (#33736)

Fynn Schmitt-Ulms committed 8 months ago

[torch.compile] Improve cold and warm start compile tests (#35709)

Richard Zou committed 8 months ago

[BugFix][Model]Fix the garbled code in Ernie4.5-VL caused by fast_moe_cold_start (#35587)

CSWYF3634076 committed 8 months ago

[CI/Build] Enable Qwen3.5 tests on CI (#35763)

Isotr0py committed 8 months ago

[CI][HPU] Pin vllm commit compatible with vllm-gaudi - HPU tests (#35307)

Patryk Wolsza committed 8 months ago

[Doc] Improve UX of `--enable-log-requests` (#35723)

Cyrus Leung committed 8 months ago

[Core] Move test utility to test file (#35672)

Turner Jabbour committed 8 months ago

[CI] Fix mypy for vllm/device allocator (#35518)

Martin Hickey committed 8 months ago

[Performance] Extract kv update ops from MLA attention backends (#34627)

ElizaWszola committed 8 months ago

[Fix Bug]`num_active_loras` always equals to zero (#34119)

Runkai Tao committed 8 months ago

[Docs] Add breadcrumbs for better UX (#35749)

Harry Mellor committed 8 months ago

[MyPy][BugFix] Check profiler is assigned before calling start() on it (#35505)

Martin Hickey committed 8 months ago