All pull requests

Pull requests

• ggml-cuda : no ring state on a ring_zone clear that has nothing to cl…

  #53·cshingle opened Nov 12, 2026·

  ·

• ci: add a single-target container build workflow

  #44·8perezm opened Nov 11, 2026·

  ·

• install : launcher defaults to ngram-simple alongside MTP

  #41·cshingle opened Nov 8, 2026·

  ·
  1

• Windows CUDA: default CUDA_MODULE_LOADING=EAGER

  #39·newmilltechnology opened Nov 7, 2026·

  ·

• Windows/MSVC: build fixes for the CPU and CUDA backends

  #38·newmilltechnology opened Nov 7, 2026·

  ·

• tests: add realistic SM80 GDN benchmark workloads

  #32·rjohny55 opened Nov 5, 2026·

  ·

• cuda: add opt-in SM80 chunked GDN prefill

  #31·rjohny55 opened Nov 5, 2026·

  ·

• ggml-cuda: fix quantized sparse attention dispatch for GQA 12

  #30·rjohny55 opened Nov 5, 2026·

  ·

• ggml-cuda : gfx1030 first-request stall

  #15·cshingle opened Nov 1, 2026·

  ·
  14