Skip to content

flashnext cuda: drafted rounds: first token early, n-gram table resident, deeper greedy drafts, CUDA graphs - #1786

Open
Arminova wants to merge 12 commits into
ashhart:mainfrom
Arminova:flashnext/draft-round
Open

Arminova wants to merge 12 commits into
ashhart:mainfrom
Arminova:flashnext/draft-round

Commits

  1. Commits on Nov 1, 2026

  2. Commits on Nov 2, 2026

  3. Commits on Nov 3, 2026