Training Language Models to Self-Correct via Reinforcement Learning

TL;DR

This paper introduces SCoRe, a reinforcement learning method enabling large language models to self-correct more effectively using only self-generated data, surpassing previous approaches in performance.

Contribution

The paper presents a novel multi-turn RL approach, SCoRe, that improves LLM self-correction without relying on multiple models or extra supervision, addressing distribution mismatch and behavior collapse issues.

Findings

SCoRe improves self-correction by 15.6% on MATH.

SCoRe improves self-correction by 9.1% on HumanEval.

State-of-the-art self-correction performance achieved.

Abstract

Self-correction is a highly desirable capability of large language models (LLMs), yet it has consistently been found to be largely ineffective in modern LLMs. Current methods for training self-correction typically depend on either multiple models, a more advanced model, or additional forms of supervision. To address these shortcomings, we develop a multi-turn online reinforcement learning (RL) approach, SCoRe, that significantly improves an LLM's self-correction ability using entirely self-generated data. To build SCoRe, we first show that variants of supervised fine-tuning (SFT) on offline model-generated correction traces are often insufficient for instilling self-correction behavior. In particular, we observe that training via SFT falls prey to either a distribution mismatch between mistakes made by the data-collection policy and the model's own responses, or to behavior collapse, where learning implicitly prefers only a certain mode of correction behavior that is often not effective at self-correction on test problems. SCoRe addresses these challenges by training under the model's own distribution of self-generated correction traces and using appropriate regularization to steer the learning process into learning a self-correction behavior that is effective at test time as opposed to fitting high-reward responses for a given prompt. This regularization process includes an initial phase of multi-turn RL on a base model to generate a policy initialization that is less susceptible to collapse, followed by using a reward bonus to amplify self-correction. With Gemini 1.0 Pro and 1.5 Flash models, we find that SCoRe achieves state-of-the-art self-correction performance, improving the base models' self-correction by 15.6% and 9.1% respectively on MATH and HumanEval.