This repository was archived by the owner on Sep 7, 2023. It is now read-only.
This repository was archived by the owner on Sep 7, 2023. It is now read-only.
New paper: "Improving Vision Transformer Efficiency and Accuracy by Learning to Tokenize" #101
Description
Detailed Description
Very interesting new paper from Google:
- Blog: Improving Vision Transformer Efficiency and Accuracy by Learning to Tokenize
- Paper: TokenLearner: What Can 8 Learned Tokens Do for Images and Videos?
Basically: A way to dynamically pick which "patches" of images / videos to attend to.
This could be a really nice way of including much larger input images (which is important for longer-time-horizon nowcasting).