A newbie question: Is it possible to partially re-use an already trained model if you decide to swap the person A?

If training is failing to start, and you are not receiving an error message telling you what to do, tell us about it here


Forum rules

Read the FAQs and search the forum before posting a new topic.

This forum is for reporting errors with the Training process. If you want to get tips, or better understand the Training process, then you should look in the Training Discussion forum.

Please mark any answers that fixed your problems so others can find the solutions.

Post Reply
User avatar
cavay
Posts: 3
Joined: Thu Mar 21, 2024 2:51 pm
Has thanked: 1 time

A newbie question: Is it possible to partially re-use an already trained model if you decide to swap the person A?

Post by cavay »

Hello,

I'm wondering if it would be somehow possible to re-use the already trained model if I decide to create a new model with the same trainer type? Initially, the model was trained with person A and person B -- now I want to keep what has been learned for person B but swap the person A.

I think this would make sense even if not currently possible.

Thanks!

User avatar
torzdf
Posts: 2667
Joined: Fri Jul 12, 2019 12:53 am
Answers: 159
Has thanked: 131 times
Been thanked: 625 times

Re: A newbie question: Is it possible to partially re-use an already trained model if you decide to swap the person A?

Post by torzdf »

You can re-use the encoder for any model with the load/freeze weights option.

You can re-use any part of an existing model with these same options with the Phaze-A model.

It is possible to load weights into any part of a new model for any faceswap model, but this functionality is not built in, and would need to be scripted outside of faceswap.

My word is final

Post Reply