One Last Attention for Your Vision-Language Model

Abstract

Pretrained vision-language models (VLMs), such as CLIP, achieve remarkable zero-shot performance, yet their downstream potential hinges on effective fine-tuning. Most adaptation methods typically focus on refining representation from separate modalities (text or vision) but neglect the critical role of their fused representations in the decision-making process, i.e., rational matrix that drives the final prediction. To bridge the gap, we propose a simple yet effective Rational Adaptaion (RAda) to explicitly exploit the final fused representation during fine-tuning. RAda employs a learned mask, obtained from a lightweight attention layer attached at the end of a VLM, to dynamically calibrate the contribution of each element in the rational matrix, enabling targeted adjustments to the final cross-modal interactions without incurring costly modifications to intermediate features. Experiments in different settings (i.e., updating, or freezing pretrained encoders in adaptation, and test-time training that can only access the unlabeled test data) show that RAda serves as a versatile fine-tuning technique, improving the baseline with minimal code and performing comparably against current arts in most settings.