STT: Soft Template Tuning for Few-Shot Adaptation

TL;DR

This paper introduces Soft Template Tuning (STT), a prompt-tuning framework that effectively adapts pre-trained models to few-shot tasks by combining manual and auto prompts, outperforming traditional fine-tuning in some cases.

Contribution

The paper proposes STT, a novel prompt-tuning method that bridges the performance gap between prompt tuning and fine-tuning in few-shot learning without extra parameters.

Findings

STT closes the performance gap between prompt tuning and fine-tuning.

STT outperforms fine-tuning on sentiment classification tasks.

STT does not require additional parameters.

Abstract

Prompt tuning has been an extremely effective tool to adapt a pre-trained model to downstream tasks. However, standard prompt-based methods mainly consider the case of sufficient data of downstream tasks. It is still unclear whether the advantage can be transferred to the few-shot regime, where only limited data are available for each downstream task. Although some works have demonstrated the potential of prompt-tuning under the few-shot setting, the main stream methods via searching discrete prompts or tuning soft prompts with limited data are still very challenging. Through extensive empirical studies, we find that there is still a gap between prompt tuning and fully fine-tuning for few-shot learning. To bridge the gap, we propose a new prompt-tuning framework, called Soft Template Tuning (STT). STT combines manual and auto prompts, and treats downstream classification tasks as a masked language modeling task. Comprehensive evaluation on different settings suggests STT can close the gap between fine-tuning and prompt-based methods without introducing additional parameters. Significantly, it can even outperform the time- and resource-consuming fine-tuning method on sentiment classification tasks.