Vision-Language Model Fine-Tuning via Simple Parameter-Efficient Modification

TL;DR

This paper introduces ClipFit, a simple parameter-efficient fine-tuning method for CLIP that only adjusts bias and normalization layers, significantly boosting zero-shot performance without extra parameters.

Contribution

It demonstrates that fine-tuning specific parameters like bias and normalization layers can effectively enhance VLMs, challenging the belief that fine-tuning degrades pre-trained knowledge.

Findings

Fine-tuning bias and normalization layers improves CLIP accuracy by 7.27%.

Low-level text bias and first normalization layer change most during fine-tuning.

ClipFit requires no additional parameters beyond existing model components.

Abstract

Recent advances in fine-tuning Vision-Language Models (VLMs) have witnessed the success of prompt tuning and adapter tuning, while the classic model fine-tuning on inherent parameters seems to be overlooked. It is believed that fine-tuning the parameters of VLMs with few-shot samples corrupts the pre-trained knowledge since fine-tuning the CLIP model even degrades performance. In this paper, we revisit this viewpoint, and propose a new perspective: fine-tuning the specific parameters instead of all will uncover the power of classic model fine-tuning on VLMs. Through our meticulous study, we propose ClipFit, a simple yet effective method to fine-tune CLIP without introducing any overhead of extra parameters. We demonstrate that by only fine-tuning the specific bias terms and normalization layers, ClipFit can improve the performance of zero-shot CLIP by 7.27\% average harmonic mean accuracy. Lastly, to understand how fine-tuning in CLIPFit affects the pre-trained models, we conducted extensive experimental analyses w.r.t. changes in internal parameters and representations. We found that low-level text bias layers and the first layer normalization layer change much more than other layers. The code is available at \url{https://github.com/minglllli/CLIPFit}.