Few-Shot Self-Rationalization with Natural Language Prompts

TL;DR

This paper explores few-shot self-rationalization in NLP, demonstrating that with proper prompting and larger models, it is possible to generate plausible explanations for predictions, though current models still lag behind human explanations.

Contribution

The paper introduces FEB, a standardized dataset collection for few-shot self-rationalization, and systematically studies prompting strategies and model scaling to improve explanation plausibility.

Findings

Scaling model size improves explanation plausibility.

Current models achieve at most 51% plausibility in explanations.

There is significant room for improvement compared to human explanations.

Abstract

Self-rationalization models that predict task labels and generate free-text elaborations for their predictions could enable more intuitive interaction with NLP systems. These models are, however, currently trained with a large amount of human-written free-text explanations for each task which hinders their broader usage. We propose to study a more realistic setting of self-rationalization using few training examples. We present FEB -- a standardized collection of four existing English-language datasets and associated metrics. We identify the right prompting approach by extensively exploring natural language prompts on FEB. Then, by using this prompt and scaling the model size, we demonstrate that making progress on few-shot self-rationalization is possible. We show there is still ample room for improvement in this task: the average plausibility of generated explanations assessed by human annotators is at most 51% (with GPT-3), while plausibility of human explanations is 76%. We hope that FEB and our proposed approach will spur the community to take on the few-shot self-rationalization challenge.