CLIP (Contrastive Language-Image Pretraining), Predict the most relevant text snippet given an image
conda install conda-forge::openai-clip