ReVeal: Self-Evolving Code Agents via Iterative Generation-Verification
ReVeal: Self-Evolving Code Agents via Iterative Generation-Verification
Recent advances in reinforcement learning (RL) with verifiable outcome rewards have significantly improved the reasoning capabilities of large language models (LLMs), especially when combined with multi-turn tool interactions. However, existing methods lack both meaningful verification signals from realistic environments and explicit optimization for verification, leading to unreliable self-verification. To address these limitations, we propose ReVeal, a multi-turn reinforcement learning framework that interleaves code generation with explicit self-verification and tool-based evaluation. ReVeal enables LLMs to autonomously generate test cases, invoke external tools for precise feedback, and improves performance via a customized RL algorithm with dense, per-turn rewards. As a result, ReVeal fosters the co-evolution of a model's generation and verification capabilities through RL training, expanding the reasoning boundaries of the base model, demonstrated by significant gains in Pass@k on LiveCodeBench. It also enables test-time scaling into deeper inference regimes, with code consistently evolving as the number of turns increases during inference, ultimately surpassing DeepSeek-R1-Zero-Qwen-32B. These findings highlight the promise of ReVeal as a scalable and effective paradigm for building more robust and autonomous AI agents.
Yiyang Jin、Kunzhao Xu、Hang Li、Xueting Han、Yanmin Zhou、Cheng Li、Jing Bai
计算技术、计算机技术
Yiyang Jin,Kunzhao Xu,Hang Li,Xueting Han,Yanmin Zhou,Cheng Li,Jing Bai.ReVeal: Self-Evolving Code Agents via Iterative Generation-Verification[EB/OL].(2025-06-12)[2025-06-28].https://arxiv.org/abs/2506.11442.点此复制
评论