Automatic Multi-Label Prompting: Simple and Interpretable Few-Shot Classification

TL;DR

AMuLaP is an automatic method for selecting label mappings in few-shot prompt-based classification, achieving competitive results without human effort or external resources.

Contribution

It introduces a simple, statistics-based algorithm for automatic multi-label prompt selection, enhancing few-shot learning with minimal manual intervention.

Findings

Achieves competitive performance on GLUE benchmark

Eliminates need for human-designed label prompts

Effective in few-shot classification scenarios

Abstract

Prompt-based learning (i.e., prompting) is an emerging paradigm for exploiting knowledge learned by a pretrained language model. In this paper, we propose Automatic Multi-Label Prompting (AMuLaP), a simple yet effective method to automatically select label mappings for few-shot text classification with prompting. Our method exploits one-to-many label mappings and a statistics-based algorithm to select label mappings given a prompt template. Our experiments demonstrate that AMuLaP achieves competitive performance on the GLUE benchmark without human effort or external resources.