SERL: Self-Examining Reinforcement Learning on Open-Domain

TL;DR

SERL introduces a self-improving reinforcement learning framework where large language models act as both the agent and evaluator, using internal reward mechanisms to enhance open-domain task performance without external signals.

Contribution

This paper presents SERL, a novel self-examining RL framework that eliminates the need for external rewards by deriving internal rewards from pairwise comparisons and self-consistency, advancing open-domain LLM capabilities.

Findings

SERL outperforms existing self-improvement methods on AlpacaEval 2.

SERL achieves state-of-the-art results among self-improving approaches.

SERL's performance is comparable to larger models like Qwen3-32B.

Abstract

Reinforcement Learning (RL) has been shown to improve the capabilities of large language models (LLMs). However, applying RL to open-domain tasks faces two key challenges: (1) the inherent subjectivity of these tasks prevents the verifiable rewards as required by Reinforcement Learning with Verifiable Rewards (RLVR); (2) Reinforcement Learning from Human Feedback (RLHF) relies on external reward mechanisms. To overcome these limitations, we propose Self-Examining Reinforcement Learning (SERL), a novel self-improving framework where the LLM serves as both Actor and Judge. SERL introduces two synergistic reward mechanisms without any external signals. On the one hand, to improve the Actor's capability, we derive rewards from Copeland-style pairwise comparison judgments across a group of generated responses. On the other hand, a self-consistency reward that encourages coherent judgments is proposed to improve the Judge's reliability. This process refines the Judge's capability, which in turn provides a more robust reward for Actor. Experiments show that our method outperforms existing self-improvement training methods. SERL improves the LC win rate of Qwen3-8B on AlpacaEval 2 from 52.37% to 59.90%. To the best of our knowledge, our method achieves state-of-the-art performance among self-improving approaches. Furthermore, it achieves a performance comparable to significantly larger models like Qwen3-32B, demonstrating superior effectiveness and robustness on open-domain tasks.