3 ms·
If you like the way NLP progresses with all the recent large models, the next natural step is to apply the same idea to video - predict the next frame. Obviousl
by p1esk 6y ago
If you like the way NLP progresses with all the recent large models, the next natural step is to apply the same idea to video - predict the next frame. Obviously, it's not the same, as we don't have a "vocabulary" of frames, so some novel approach is needed. I haven't actually looked into video prediction/generation literature, so I don't know what's happening in that field.
- jonathanbesomi 6y agoInteresting; thank you! Related to videos: recently Open AI released GPT-3, a transformer models trained this time on images ...
- p1esk 6y agoI just did a quick search, here are a few relevant papers: https://arxiv.org/abs/2006.10704 https://arxiv.org/abs/2006.10704 https://arxiv.org/abs/1903.00271 https://arxiv.org/abs/1903.00271 http://www.cs.columbia.edu/~vondrick/transformer.pdf http://www.cs.columbia.edu/~vondrick/transformer.pdf