Svelte Hacker News logo
  • top
  • new
  • show
  • ask
  • jobs
  • about

Batched reward model inference and Best-of-N sampling

raw.sh

33 points by rawsh 4 days ago