Classify images using a pruned Vision Transformer model
Semantic image retrieval with zero-shot classification