Distilling Out-of-Distribution Robustness from Vision-Language Foundation Models | Read Paper on Bytez