8000 Pull requests · marin-community/vllm · GitHub
[go: up one dir, main page]
Skip to content

Repository navigation

All pull requests

    Pull requests

    Open
     (11)
    Closed
     (96)
    #115Helw150 opened Nov 3, 2026Member

  • Use native GQA for GrugMoE KV caches

    #110yonromai opened Nov 1, 2026Member
    1

  • Write ShortConv results into its output buffer

    #107yonromai opened Oct 29, 2026Member

  • [release] Bundle the source tests for GPU wheel qualification

    #106atqamar opened Oct 26, 2026Member

  • [GrugMoE] Add an opt-in row-invariant GEMM for the router and head gate

    #96atqamar opened Oct 18, 2026Member

  • Remove obsolete GPU compatibility overrides

    #95yonromai opened Oct 14, 2026Member

  • Handle omitted Grug short-convolution metadata during profiling

    #83dlwh opened Oct 2, 2026Member
    1

  • Link Marin serving tasks to the performance guides

    #82yonromai opened Sep 30, 2026Member
    1
    •
    0