Abstract: Prompt learning stands out as one of the most efficient approaches for adapting powerful vision-language foundational models like CLIP to downstream datasets by tuning learnable prompt ...