[Docs] Add ECMooncakeConnector usage example for EPD (#57742)

jiangkuaixue123 committed about 1 month ago

[Bugfix][Spec Decode] Stop dummy draft decode steps from writing KV through stale block-table rows (#56734)

Yifan Qiao committed about 1 month ago

[Perf][DSV4.1] Overlap mHC coefficients for small TP batches (#57603)

Woosuk Kwon committed about 1 month ago

[XPU][UT]Bugfix when the process can't see all the world_size meet accuracy issue. (#57779)

YiSheng5 committed about 1 month ago

[KVConnector][MoRIIO] Transfer hybrid mamba/KDA recurrent state in READ mode (#51052)

YukioZzz committed about 1 month ago

[ROCm][CI] Query HIP device memory for test GPU teardown waits. (#57450)

Sheral Kumar committed about 1 month ago

[Refactor] Remove dead kernel code (#57621)

Wentao Ye committed about 1 month ago

[Misc] Rename --enable-mamba-fine-grained-prefix-cache (#53945 follow-up) (#57382)

roikoren755 committed about 1 month ago

[Feature][Humming] Humming feature integration (#56685)

Jinzhen Lin committed about 1 month ago

[DSV4.1] Add encoder cuda graph support for deepseek-v4.1-flash (#56625)

Isotr0py committed about 1 month ago

[Feature] support bf16 MoE router and mxfp4 MoE for MiMo V2 (#57784)

Zyann committed about 1 month ago

[Frontend] Expose per-request spec decode metrics in generate API (#43310)

Hongxin Xu committed about 1 month ago

[GLM-5.3-Flash] Route kpool indexer top-k through the shared SparseIndexerTopk dispatcher (#57546)

Chauncey committed about 1 month ago

[Bugfix][KV Offload] Skip non-prefix-cacheable groups in SimpleCPUOffload (GLM-5.3-Flash kpool tail and QSA) (#56810)

Jared Wen committed about 1 month ago

[Perf][GLM] Fuse the kpool tail slot mapping into one Triton kernel (#57534)

Jared Wen committed about 1 month ago

[Bugfix][GLM-5.3-Flash] Address kpool tail blocks by the padded indexer stride in the NVIDIA prefill seed kernel (#57477)

Jared Wen committed about 1 month ago

[Model][LoRA] Enable LoRA support for VoyageQwen3BidirectionalEmbedModel (#57708)

mahird3 committed about 1 month ago

[Core][Kernel] Share persistent workspaces for Marlin and Humming (#57421)

Misha Goin committed about 1 month ago

[ROCm][DSV4][Perf] Use FP8 WO_A output projection (#54894)

yinfengLiu committed about 1 month ago

[Bugfix][Frontend] Bound the prompt after multimodal expansion (#57076)

Dilber committed about 1 month ago