See the contents of the accompanying Jupyter notebook for a simple implementation of Chefer et al. (2021) that works with CLIP-ViT via the Hugging Face API
morrisalp / clip-explainability Goto Github PK
View Code? Open in Web Editor NEWA simple demonstration of explainable image-text similarity with CLIP-ViT, based on Chefer et al. (ICCV 2021)