DePT: Decoupled Prompt Tuning

TL;DR

DePT introduces a method to decouple base-specific knowledge from shared features during prompt tuning, significantly improving zero-shot generalization to new tasks by addressing the channel bias issue.

Contribution

The paper proposes Decoupled Prompt Tuning (DePT), a novel framework that isolates base-specific knowledge to enhance task-shared feature preservation and generalization.

Findings

DePT improves zero-shot performance across 11 datasets.

Decoupling base-specific knowledge enhances prompt tuning flexibility.

DePT is compatible with existing prompt tuning methods.

Abstract

This work breaks through the Base-New Tradeoff (BNT)dilemma in prompt tuning, i.e., the better the tuned model generalizes to the base (or target) task, the worse it generalizes to new tasks, and vice versa. Specifically, through an in-depth analysis of the learned features of the base and new tasks, we observe that the BNT stems from a channel bias issue, i.e., the vast majority of feature channels are occupied by base-specific knowledge, resulting in the collapse of taskshared knowledge important to new tasks. To address this, we propose the Decoupled Prompt Tuning (DePT) framework, which decouples base-specific knowledge from feature channels into an isolated feature space during prompt tuning, so as to maximally preserve task-shared knowledge in the original feature space for achieving better zero-shot generalization on new tasks. Importantly, our DePT is orthogonal to existing prompt tuning methods, hence it can improve all of them. Extensive experiments on 11 datasets show the strong flexibility and effectiveness of DePT. Our code and pretrained models are available at https://github.com/Koorye/DePT.