Improving Cross-Task Generalization with Step-by-Step Instructions

TL;DR

This paper introduces a method to improve cross-task generalization in language models by automatically generating and incorporating step-by-step instructions, which decompose tasks into detailed procedures, enhancing model performance.

Contribution

The paper proposes automatically generating step-by-step instructions via ChatGPT and combining them with original instructions to improve task decomposition and generalization in language models.

Findings

Step-by-step instructions improve cross-task generalization.

Order of steps significantly affects performance.

High-quality instructions lead to better results.

Abstract

Instruction tuning has been shown to be able to improve cross-task generalization of language models. However, it is still challenging for language models to complete the target tasks following the instructions, as the instructions are general and lack intermediate steps. To address this problem, we propose to incorporate the step-by-step instructions to help language models to decompose the tasks, which can provide the detailed and specific procedures for completing the target tasks. The step-by-step instructions are obtained automatically by prompting ChatGPT, which are further combined with the original instructions to tune language models. The extensive experiments on SUP-NATINST show that the high-quality step-by-step instructions can improve cross-task generalization across different model sizes. Moreover, the further analysis indicates the importance of the order of steps of the step-by-step instruction for the improvement. To facilitate future research, we release the step-by-step instructions and their human quality evaluation results.