Preserving Multi-Modal Capabilities of Pre-trained VLMs for Improving Vision-Linguistic Compositionality

TL;DR

This paper introduces FSC-CLIP, a method that enhances compositional understanding in pre-trained vision-language models while maintaining their multi-modal capabilities, addressing limitations of traditional fine-tuning approaches.

Contribution

FSC-CLIP integrates local hard negative loss and selective regularization to improve compositionality without degrading multi-modal performance.

Findings

Achieves state-of-the-art compositionality performance

Retains strong multi-modal capabilities

Effective across diverse benchmarks

Abstract

In this paper, we propose a new method to enhance compositional understanding in pre-trained vision and language models (VLMs) without sacrificing performance in zero-shot multi-modal tasks. Traditional fine-tuning approaches often improve compositional reasoning at the cost of degrading multi-modal capabilities, primarily due to the use of global hard negative (HN) loss, which contrasts global representations of images and texts. This global HN loss pushes HN texts that are highly similar to the original ones, damaging the model's multi-modal representations. To overcome this limitation, we propose Fine-grained Selective Calibrated CLIP (FSC-CLIP), which integrates local hard negative loss and selective calibrated regularization. These innovations provide fine-grained negative supervision while preserving the model's representational integrity. Our extensive evaluations across diverse benchmarks for both compositionality and multi-modal tasks show that FSC-CLIP not only achieves compositionality on par with state-of-the-art models but also retains strong multi-modal capabilities. Code is available at: https://github.com/ytaek-oh/fsc-clip.