让大模型先反思再行动，显著减少错误累积。
Reflect before Act: Proactive Error Correction in Language Models
- 在执行动作前增加反思步骤，主动发现并修正错误。
- 在WebShop等环境中成功率最高提升24%，达61%。
- 仅需少量修改即可实现，适合实时交互系统部署。
大语言模型在交互式决策任务中表现出色，但现有方法常因错误累积而失效，且缺乏可靠的自我修正能力。本文提出「反思再行动」（Reflect before Act, REBACT），通过在每一步动作前引入反思环节，实现即时错误纠正，确保行动路径流畅并适应环境反馈。我们在ALFWorld、WebShop和TextCraft三个不同交互环境上评估REBACT，结果表明其显著优于强基线：在WebShop上成功率提升24%（达61%），在ALFWorld上提升6.72%（达98.51%），在TextCraft上提升0.5%（达99.5%），均基于Claude3.5-sonnet模型。进一步分析显示，性能提升仅需少量修改步骤，具备良好计算效率。
原文摘要 · Abstract (English)
Large Language Models (LLMs) have demonstrated remarkable capabilities in interactive decision-making tasks, but existing methods often struggle with error accumulation and lack robust self-correction mechanisms. We introduce "Reflect before Act" (REBACT), a novel approach that enhances LLM-based decision-making by introducing a critical reflect step prior to taking the next action. This approach allows for immediate error correction, ensuring smooth action path and adaptibity to environment feedback. We evaluate REBACT on three diverse interactive environments: ALFWorld, WebShop, and TextCraft. Our results demonstrate that REBACT significantly outperforms strong baselines, improving success rates by up to 24% on WebShop (achieving 61%), 6.72% on ALFWorld (achieving 98.51%), and 0.5% on TextCraft (achieving 99.5%) using Claude3.5-sonnet as the underlying LLM. Further analysis reveals that REBACT's performance improvements are achieved with only a few modification steps, demonstrating its computational efficiency.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。