causal

RL Unlocks Skill, Pre-Training Provides Knowledge

RL post-training does not teach the model new knowledge but unlocks the knowledge it already has from pre-training—it is skill learning (how to solve problems) rather than knowledge soaking, which is why RLVR can take a base model from 15% to 50% accuracy on MATH-500 in 50 steps without teaching new math.

causalpending

Speaker

Sebastian Raschka

Evidence Quote

“the knowledge is already there in the pre-training; you’re just unlocking it.”

Source

Ai Sota #2026— Lex Fridman Podcast
Created: 6/13/2026, 3:24:41 AM

Your notes

Loading notes...