Leap-Of-Thought: Teaching Pre-Trained Models to Systematically Reason Over Implicit Knowledge
Alon Talmor1,2 Oyvind Tafjord1 Peter Clark1 Yoav Goldberg1,3 Jonathan Berant1,2 1The Allen Institute for AI 2Tel-Aviv University, 3Bar-Ilan University
Abstract
To what extent can a neural network systematically reason over symbolic facts? Evidence suggests that large pre-trained language models (LMs) acquire some reasoning capacity, but this ability is difficult to control. Recently, it has been shown that Transformer-based models succeed in consistent reasoning over explicit symbolic facts, under a “closed-world" assumption. However, in an open-domain setup, it is desirable to tap into the vast reservoir of implicit knowledge already encoded in the parameters of pre-trained LMs. In this work, we provide a first demonstration that LMs can be trained to reliably perform systematic reasoning combining both implicit, pre-trained knowledge and explicit natural language statements. To do this, we describe a procedure for automatically generating datasets that teach a model new reasoning skills, and demonstrate that models learn to effectively perform inference which involves implicit taxonomic and world knowledge, chaining and counting. Finally, we show that “teaching” the models to reason generalizes beyond the training distribution: they successfully compose the usage of multiple reasoning skills in single examples. Our work paves a path tow
中文速览
大型语言模型（LM）虽然在预训练中积累了大量隐性世界知识，但如何让它把这些"藏在参数里的知识"和用户临时提供的显式自然语言陈述结合起来进行可靠推理，一直是个难题。研究者提出了一套自动数据生成流程，利用 ConceptNet、WordNet、Wikidata 等知识库构造训练样本，专门教会 RoBERTa 等预训练模型学会"分类继承推理"（如知道鲸鱼是哺乳动物，再结合用户给出的"哺乳动物有肚脐"就能推出鲸鱼有肚脐）和"计数推理"两类技能。实验表明，经过这种训练的模型不仅能在测试时准确融合隐性与显性知识，还能把分别学到的多种推理技能自发组合用于训练分布之外的新例子，展现出良好的泛化能力。这项工作意义在于：它为构建"随用随更新"的开放域系统铺平了道路——用户只需提供一句自然语言纠正，模型就能借助已有推理能力立即修正错误，无需重新收集数据或重新训练。
原文 arXiv:2006.06609;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2006.06609v3