Consistent Self-Attention for Long-Range Image and Video Generation (Updated 1 year, 1 month ago)
This model is cold. You'll get a fast response if the model is warm and already running, and a slower response if the model is cold and starting up.