Leveraging the Inductive Bias of Large Language Models for Abstract Textual Reasoning

TL;DR

This paper explores how large language models' inherent inductive biases can be leveraged to improve abstract textual reasoning, enabling better generalization and compositional learning in symbolic reasoning tasks.

Contribution

It demonstrates that large language models can serve as effective inductive biases for training symbolic reasoning engines, enhancing their generalization and learning efficiency.

Findings

Models generalize naturally to new scenarios.

Training on simpler tasks improves performance on complex tasks.

Language captures structural knowledge aiding reasoning.

Abstract

Large natural language models (such as GPT-3 or T5) demonstrate impressive abilities across a range of general NLP tasks. Here, we show that the knowledge embedded in such models provides a useful inductive bias, not just on traditional NLP tasks, but also in the nontraditional task of training a symbolic reasoning engine. We observe that these engines learn quickly and generalize in a natural way that reflects human intuition. For example, training such a system to model block-stacking might naturally generalize to stacking other types of objects because of structure in the real world that has been partially captured by the language describing it. We study several abstract textual reasoning tasks, such as object manipulation and navigation, and demonstrate multiple types of generalization to novel scenarios and the symbols that comprise them. We also demonstrate the surprising utility of \textit{compositional learning}, where a learner dedicated to mastering a complicated task gains an advantage by training on relevant simpler tasks instead of jumping straight to the complicated task.