[Bug] Fix TypeError when hf_config.architectures is None during model loading (#38849)

Tihomir Elek committed 7 months ago

[MXFP8] [XPU] add a new compressed tensor schema and add a xpu mxfp8 gemm kernel (#38707)

zofia committed 7 months ago

[Bugfix] Fix Responses API instructions leaking through previous_response_id (#37727)

Yufeng He committed 7 months ago

feat: add max_tokens_per_doc in rerank request. (#38827)

Jesus Federico committed 7 months ago

[Bugfix] Fix GLM tool parser streaming with MTP or stream interval (#39253)

Flora Feng committed 7 months ago

[XPU]Enhance environment collection for Intel XPU and optimize layout (#35698)

sihao_li committed 7 months ago

feat: rename logit_bias/logit_scale to logit_mean/logit_sigma for affine score calibration (#39530)

Jesus Federico committed 7 months ago

fix(lmcache): correct store for cached requests and num_scheduled_tokens in lmcache_mp_connector.py (#39655)

maobaolong committed 7 months ago

fused qknorm+rope kernel optimization for SM9.0 (#37376)

Kaicheng Yang committed 7 months ago

[compile] Enable AOT compile with batch invariance mode. (#39201)

Zhengxu Chen committed 7 months ago

[LoRA] Support dual CUDA streams-Linear Layer (#35721)

Jee Jee Li committed 7 months ago

[XPU] revert torch-xpu to 2.10 (#39656)

Kunshang Ji committed 7 months ago

[ROCm][CI] Removed stale tests and extended acceptance test (#39651)

Andreas Karatzas committed 7 months ago

[Bugfix] [Tests] Enforce `out` tensor device in `kernel/moe/test_cutedsl_moe.py` (#39644)

Yongye Zhu committed 7 months ago

[compile] Bug fix for _decompose_size_nodes (#38360)

Animesh Jain committed 7 months ago

fix(moe): fix RoutedExpertsCapturer assertion failure with DP>1 and MK path (#37879)

Le Yang committed 7 months ago

[KVConnector][NIXL] Organize NIXL connector into its own directory (#39354)

Nicolò Lucchesi committed 7 months ago

[Core][Metrics] Remove `vllm:prompt_tokens_recomputed` metric (#38709)

Mark McLoughlin committed 7 months ago

[Pooling] Disable async scheduling by default for pooling models (#39592)

Nick Hill committed 7 months ago

[HMA] [KVEvent] Enable GPU-side KV events for HMA (#37688)

Martin Hickey committed 7 months ago