Skip to content

cuda: Bonsai 2 27B at the full 262k window with the MTP head on 12 GB Ada: #218 + #215 hybrid PTQ1_0 dispatch, in-place q4_0/q8_0 K/V flash attention (integration checkpoint) - #221

Merged
bri-prism merged 27 commits into
PrismML-Eng:prismfrom
professorpalmer:bonsai-combo
Sep 29, 2026
Merged

cuda: Bonsai 2 27B at the full 262k window with the MTP head on 12 GB Ada: #218 + #215 hybrid PTQ1_0 dispatch, in-place q4_0/q8_0 K/V flash attention (integration checkpoint)#221
bri-prism merged 27 commits into
PrismML-Eng:prismfrom
professorpalmer:bonsai-combo

Commits

  1. Commits on Sep 25, 2026

  2. Commits on Sep 26, 2026