Reply To: Can i train same model?

Home Forums DeepFaceLab Training Can i train same model? Reply To: Can i train same model?

#7516
deepfakery
Keymaster

    Yes you can! In general the swap information is held in the model file(s) named with “inter”. When starting a new project with the same model you want to delete the inter files and go back to the random warp phase. LIAE has 2 inter files: inter_B is the destination info and inter_AB is the predicated face info. So, if you’re using the same source face again you can delete only inter_B. However if you are using both a new src and dst then you should delete both inter files.