SuperPos-Prompt: Enhancing Soft Prompt Tuning of Language Models with Superposition of Multi Token Embeddings

TL;DR

SuperPos-Prompt introduces a superposition-based reparameterization for soft prompt tuning, significantly improving performance and convergence speed on benchmark tasks, sometimes surpassing full fine-tuning.

Contribution

It proposes a novel superposition technique for soft prompt tuning and shows its effectiveness across multiple benchmarks, outperforming existing methods.

Findings

SuperPos-Prompt outperforms Residual Prompt tuning with +6.4 and +5.0 score improvements.

Omitting dropouts enhances performance and convergence speed.

SuperPos-Prompt sometimes surpasses full fine-tuning methods.

Abstract

Soft prompt tuning techniques have recently gained traction as an effective strategy for the parameter-efficient tuning of pretrained language models, particularly minimizing the required adjustment of model parameters. Despite their growing use, achieving optimal tuning with soft prompts, especially for smaller datasets, remains a substantial challenge. This study makes two contributions in this domain: (i) we introduce SuperPos-Prompt, a new reparameterization technique employing the superposition of multiple pretrained vocabulary embeddings to improve the learning of soft prompts. Our experiments across several GLUE and SuperGLUE benchmarks consistently highlight SuperPos-Prompt's superiority over Residual Prompt tuning, exhibiting an average score increase of $+6.4$ in T5-Small and $+5.0$ in T5-Base along with a faster convergence. Remarkably, SuperPos-Prompt occasionally outperforms even full fine-tuning methods. (ii) Additionally, we demonstrate enhanced performance and rapid convergence by omitting dropouts from the frozen network, yielding consistent improvements across various scenarios and tuning methods.