The currently released preview model was developed based on Cosmos-Predict2-2B-Text2Image. Will the final model be further trained on this preview model, or will training proceed using a different base model?
· Sign up or log in to comment
