arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2603.02091cs.LGcs.AIcs.CL
收藏

通过合成数据学习提升多跳推理能力

Learning from Synthetic Data Improves Multi-hop Reasoning

Cornell University(康奈尔大学);University of Cambridge(剑桥大学);Stanford University(斯坦福大学)

Anmol Kabra, Yilun Yin, Albert Gong, Kamilė Stankevičiūtė, Dongyoung Go, Johann Lee, Katie Z. Luo, Carla P. Gomes, Kilian Q. Weinberger

AI总结:

本研究通过规则生成的合成数据提升LLM多跳推理能力,发现合成数据能有效训练模型组成知识,从而在现实问答任务中表现更优。

摘要:

强化学习(RL)已被证明能显著提升大语言模型(LLMs)在数学、编程和多跳推理任务中的推理能力。然而,RL微调需要大量高质量可验证数据,通常来源于人工标注、前沿LLM生成或LLM基于的验证器评分。这三者都有显著局限:人工标注的数据集较小且成本高,LLM生成的数据容易产生幻觉且成本高,LLM基于的验证器不准确且速度慢。在本工作中,我们探讨了一种更经济的替代方法:在规则生成的合成数据上进行多跳推理任务的RL微调。我们发现,经过合成数据微调的LLMs在流行的现实世界问答基准测试中表现显著更好,尽管合成数据仅包含虚构知识。在按问题难度分层性能时,我们发现合成数据教会LLMs组成知识——一种基本且可推广的推理技能。我们的工作强调了规则生成的合成推理数据作为免费且可扩展的资源,以提升LLM的推理能力。

英文摘要:

Reinforcement Learning (RL) has been shown to significantly boost reasoning capabilities of large language models (LLMs) in math, coding, and multi-hop reasoning tasks. However, RL fine-tuning requires abundant high-quality verifiable data, often sourced from human annotations, generated from frontier LLMs, or scored by LLM-based verifiers. All three have considerable limitations: human-annotated datasets are small and expensive to curate, LLM-generated data is hallucination-prone and costly, and LLM-based verifiers are inaccurate and slow. In this work, we investigate a cheaper alternative: RL fine-tuning on rule-generated synthetic data for multi-hop reasoning tasks. We discover that LLMs fine-tuned on synthetic data perform significantly better on popular real-world question-answering benchmarks, despite the synthetic data containing only fictional knowledge. On stratifying performance by question difficulty, we find that synthetic data teaches LLMs to compose knowledge -- a fundamental and generalizable reasoning skill. Our work highlights rule-generated synthetic reasoning data as a free and scalable resource to improve LLM reasoning capabilities.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑