Quantifying Interpretability in CLIP Models with Concept Consistency

VLM
Re-assign community