factual
Reasoning behaviors emerge from RL without human examples
DeepSeek-R1-Zero showed that reasoning behaviors (backtracking, self-checking, 'wait let me verify') emerge naturally from large-scale reinforcement learning on verifiable questions with no human preference data or human-written reasoning chains added, analogous to AlphaZero removing the human prior to become more powerful.
factualpending
Speaker
Nathan LambertEvidence Quote
“these reasoning behaviors emerge naturally. So these things like, “Wait, let me see... Oh, that might be a mistake.” And they emerge from only having questions and answers.”
Created: 6/13/2026, 3:31:25 AM
Your notes
Loading notes...