Comment

Comment by robrenaud

Do you understand why RL is better than SFT for training on reasoning traces?
parent robrenaud 1 year ago

Discussion

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton