serve: constrained requests speculate, and speculative decode falls back to plain by default (refs #1781) - #2297

serve: constrained requests speculate, and speculative decode falls back to plain by default (refs #1781)#2297

bong-water-water-bong merged 3 commits into

main1bit-MONSTER/engine:mainfrom

cl-npuperf/1781-spec-grammar21bit-MONSTER/engine:cl-npuperf/1781-spec-grammar2Copy head branch name to clipboard