arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.03268cs.LGcs.AIcs.CL
收藏

从推理路径聚合视角理解语言模型的推理能力

Understanding Reasoning Ability of Language Models From the Perspective of Reasoning Paths Aggregation

University of California, Santa Barbara(加利福尼亚大学圣巴巴拉分校);Carnegie Mellon University(卡内基梅隆大学);University of Waterloo(滑铁卢大学)

Xinyi Wang, Alfonso Amayuelas, Kexun Zhang, Liangming Pan, Wenhu Chen, William Yang Wang

AI总结:

该研究从推理路径聚合视角解释语言模型的推理涌现机制,将推理路径形式化为图上随机游走路径,验证了该视角的有效性并提出路径扩充方法以提升多步推理性能。

摘要:

预训练语言模型(LM)无需显式微调即可执行复杂推理。为理解采用下一词元预测目标的预训练如何促成这种推理能力的涌现,我们提出可将LM视为通过聚合预训练阶段见过的间接推理路径来推导新结论。我们发现该视角在两类重要推理场景中有效:基于知识图谱(KG)的逻辑推理与思维链(CoT)推理。具体而言,我们将推理路径形式化为知识/推理图上的随机游走路径。对LM习得分布的分析表明,相关随机游走路径概率的加权和是解释LM推理机制的合理方式。在多个KG和CoT数据集上的实验与分析揭示了在随机游走路径上训练的效果,并表明扩充无标注随机游走推理路径可提升真实世界多步推理性能。代码:https://github.com/WANGXinyiLinda/LM_random_walk

英文摘要:

Pre-trained language models (LMs) are able to perform complex reasoning without explicit fine-tuning. To understand how pre-training with a next-token prediction objective contributes to the emergence of such reasoning capability, we propose that we can view an LM as deriving new conclusions by aggregating indirect reasoning paths seen at pre-training time. We found this perspective effective in two important cases of reasoning: logic reasoning with knowledge graphs (KGs) and chain-of-thought (CoT) reasoning. More specifically, we formalize the reasoning paths as random walk paths on the knowledge/reasoning graphs. Analyses of learned LM distributions suggest that a weighted sum of relevant random walk path probabilities is a reasonable way to explain how LMs reason. Experiments and analysis on multiple KG and CoT datasets reveal the effect of training on random walk paths and suggest that augmenting unlabeled random walk reasoning paths can improve real-world multi-step reasoning performance. code: https://github.com/WANGXinyiLinda/LM_random_walk

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。

↑