Hello, thank you for sharing the great work.
I wonder when the code will be released.
Also, I have a question about the intuition of using an unmasked teacher in UMT and InternVideo2.
Why do you use feature distillation rather than directly adopting MAE-family models?
Hello, thank you for sharing the great work. I wonder when the code will be released. Also, I have a question about the intuition of using an unmasked teacher in UMT and InternVideo2. Why do you use feature distillation rather than directly adopting MAE-family models?