You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am thinking of extending your works further. But there's one thing I don't understand clearly from your paper and implementation.
In your paper Figure 3 (the main figure of the procedure), in the Forward part, why is c going to 'Regularized Forward DDIM' as input? Doesn't DDIM forward process usually only adds noise so text condition c is not needed in the process? Can you clarify this part?
Thank you.
The text was updated successfully, but these errors were encountered:
Thank you for your interest in this project!
The text conditioning c is going to the Regularized Forward DDIM as input because it is needed to compute how much noise needs to be added (as also shown in equation 1 in our paper).
Thank you for great works!
I am thinking of extending your works further. But there's one thing I don't understand clearly from your paper and implementation.
In your paper Figure 3 (the main figure of the procedure), in the Forward part, why is c going to 'Regularized Forward DDIM' as input? Doesn't DDIM forward process usually only adds noise so text condition c is not needed in the process? Can you clarify this part?
Thank you.
The text was updated successfully, but these errors were encountered: