5fa1a76
1
After pretraining, you can use natural language to instruct CLIP to predict the text given an image or vice versa.
