arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.01993cs.CLcs.AI
收藏

SAFE: 一种基于LLM作为验证器的证据驱动多跳推理框架

SAFE: An LLM-as-Verifier Framework for Evidence-Grounded Multi-Hop Reasoning

Seoul National University(首尔国立大学)

Daeyong Kwon, Soyoung Yoon, Seung-won Hwang

AI总结:

提出SAFE框架,通过将推理分解为知识图谱三元组,在生成过程中逐步验证中间步骤,以解决多跳问答中模型通过无效推理得到正确答案的问题,平均准确率提升8.8个百分点。

摘要:

多跳问答基准测试常常奖励大型语言模型(LLM)的虚假正确性,即模型通过无效的中间推理得出正确答案。我们提出了SAFE,一种基于LLM作为验证器的证据驱动多跳问答框架。SAFE不是在生成后仅判断最终答案,而是在生成过程中通过检查中间步骤与提供的段落和先前的推理轨迹来验证推理。为了使这一过程可检查,SAFE将推理分解为以知识图谱(KG)三元组表示的原子化、证据驱动的单元。在训练时,SAFE在KG约束下验证基准监督,并构建可靠的验证器训练数据。在推理时,外部验证器检查每个生成的步骤,识别无效推理,并在错误传播之前提供纠正反馈。在三个多跳问答基准测试中,SAFE平均提高了8.8个百分点的准确率。这些结果表明,证据驱动的多跳问答受益于将基于LLM的评估从事后答案判断转向逐步推理验证。

英文摘要:

Multi-hop QA benchmarks often reward Large Language Models (LLMs) for spurious correctness, where models reach correct answers through invalid intermediate reasoning. We propose SAFE, an LLM-as-verifier framework for evidence-grounded multi-hop QA. Rather than judging only the final answer after generation, SAFE verifies reasoning during generation by checking intermediate steps against the provided passages and previous reasoning trajectory. To make this process checkable, SAFE decomposes reasoning into atomic, evidence-grounded units represented with Knowledge Graph (KG) triples. At train-time, SAFE verifies benchmark supervision under KG-grounded constraints and constructs reliable verifier training data. At inference-time, an external verifier checks each generated step, identifies invalid reasoning, and provides correction feedback before errors propagate. Across three multi-hop QA benchmarks, SAFE improves accuracy by 8.8 pp on average. These results show that evidence-grounded multi-hop QA benefits from shifting LLM-based evaluation from post-hoc answer judgment to stepwise reasoning verification.

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。

↑