Leveraging Vision-Language Models for Domain Generalization in Image Classification

The author proposes a method, VL2V-ADiP, to distill Vision-Language Models for better Out-of-Distribution generalization in image classification tasks.