Back to Basics: Revisiting Exploration in Reinforcement Learning for LLM Reasoning via Generative Probabilities

TL;DR

This paper introduces a novel advantage re-weighting mechanism to improve exploration and diversity in reinforcement learning for large language models, addressing mode collapse and entropy reduction.

Contribution

It proposes a new advantage re-weighting method that balances confidence levels across responses, enhancing diversity without sacrificing accuracy in LLM reasoning tasks.

Findings

Significantly increases generative diversity and response entropy.

Outperforms existing methods by 5.7% in Pass@1 and 13.9% in Pass@32 on Qwen2.5-7B.

Effectively mitigates entropy collapse in reinforcement learning for LLMs.

Abstract

Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as an indispensable paradigm for enhancing reasoning in Large Language Models (LLMs). However, standard policy optimization methods, such as Group Relative Policy Optimization (GRPO), often converge to low-entropy policies, leading to severe mode collapse and limited output diversity. We analyze this issue from the perspective of sampling probability dynamics, identifying that the standard objective disproportionately reinforces the highest-likelihood paths, thereby suppressing valid alternative reasoning chains. To address this, we propose a novel Advantage Re-weighting Mechanism (ARM) designed to equilibrate the confidence levels across all correct responses. By incorporating Prompt Perplexity and Answer Confidence into the advantage estimation, our method dynamically reshapes the reward signal to attenuate the gradient updates of over-confident reasoning paths, while redistributing probability mass toward under-explored correct solutions. Empirical results demonstrate that our approach significantly enhances generative diversity and response entropy while maintaining competitive accuracy, effectively achieving a superior trade-off between exploration and exploitation in reasoning tasks. Empirical results on Qwen2.5 and DeepSeek models across mathematical and coding benchmarks show that ProGRPO significantly mitigates entropy collapse. Specifically, on Qwen2.5-7B, our method outperforms GRPO by 5.7% in Pass@1 and, notably, by 13.9% in Pass@32, highlighting its superior capability in generating diverse correct reasoning paths.