Nemesis: Normalizing the Soft-prompt Vectors of Vision-Language Models

TL;DR

Nemesis introduces a normalization technique for soft prompts in vision-language models, revealing that adjusting prompt norms can improve model performance and offering new insights into prompt tuning strategies.

Contribution

This work is the first to systematically analyze the impact of soft prompt norms in VLMs and proposes a normalization method called Nemesis to enhance their performance.

Findings

Reducing prompt norms can improve VLM performance.

Increasing prompt norms often degrades model accuracy.

Normalization of soft prompts leads to better downstream task results.

Abstract

With the prevalence of large-scale pretrained vision-language models (VLMs), such as CLIP, soft-prompt tuning has become a popular method for adapting these models to various downstream tasks. However, few works delve into the inherent properties of learnable soft-prompt vectors, specifically the impact of their norms to the performance of VLMs. This motivates us to pose an unexplored research question: ``Do we need to normalize the soft prompts in VLMs?'' To fill this research gap, we first uncover a phenomenon, called the \textbf{Low-Norm Effect} by performing extensive corruption experiments, suggesting that reducing the norms of certain learned prompts occasionally enhances the performance of VLMs, while increasing them often degrades it. To harness this effect, we propose a novel method named \textbf{N}ormalizing th\textbf{e} soft-pro\textbf{m}pt v\textbf{e}ctors of vi\textbf{si}on-language model\textbf{s} (\textbf{Nemesis}) to normalize soft-prompt vectors in VLMs. To the best of our knowledge, our work is the first to systematically investigate the role of norms of soft-prompt vector in VLMs, offering valuable insights for future research in soft-prompt tuning. The code is available at \texttt{\href{https://github.com/ShyFoo/Nemesis}{https://github.com/ShyFoo/Nemesis}}.