Skip to content

server/llama: honor per-request speculative.n_max, per-sequence snapshot budget, honest rollback refusal (multi-slot decode with MTP) - #208

Open
alenizy361 wants to merge 15 commits into
PrismML-Eng:prismfrom
alenizy361:server-spec-multislot
Open

alenizy361 wants to merge 15 commits into
PrismML-Eng:prismfrom
alenizy361:server-spec-multislot

Commits

  1. Commits on Sep 19, 2026