fix: take multiple style reference images from one_style.json#29
Open
pinkfloat wants to merge 1 commit into
Open
fix: take multiple style reference images from one_style.json#29pinkfloat wants to merge 1 commit into
pinkfloat wants to merge 1 commit into
Conversation
The previous reference to index 0 took only the first style reference listed in one_style.json to train on it.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Hi, hope I don't bother you.
I was using your cool implementation for a recent university project of mine where I compare a few style-adjustment architectures - anyways.
At some point I noticed, that I was getting the exact same training results for putting only one description into
one_style.json- like this:or multiple - as long as the first image was the same - like this:
And I kinda noticed that late ... like: At my twentieth trainings-run or so. At this point I already had started to believe that the model always tends to over-fit on specific textures and the results simply never get as good for me as in the provided examples.
Which kinda makes sense, if I always overfit on exactly one training image - my first. Because all the other listed images in one_style.json are silently ignored. So ... changing these lines helped me to (finally!) get good results at the end. And this fix is kinda not very obvious.
So maybe you think about it.
Cheers~