[bug fixes - advanced sdxl lora script] - add left lora changes from canonical script and fix bug in snr gamma #6327
Conversation
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
Collaborator
Author
|
cc @apolinario @sayakpaul for review 🙏 |
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
Member
|
Should #6294 be closed? |
Collaborator
Author
my bad! This one should only be the snr gamma - prior preservation bug fix then, I'll close this and open a new one for the bug fix. Does the change for the snr gamma calculation make sense logically do you think? |
Collaborator
Author
|
closing to not overlap with #6294, will open a separate one for the snr gamma bug |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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Outdated suggestions cannot be applied.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.
This PR updates and fixes issues in the advanced dreambooth lora sdxl training script
min SNR gammaandprior preservation loss:when using both, there is a shape mismatch between the instance
lossand themse_loss_weightswhen trying to perform the following operation:loss = loss.mean(dim=list(range(1, len(loss.shape)))) * mse_loss_weightscausing an error.
I think the reason is the shape of
timestepsthat is used to calculate the snr gamma - we use timesteps sampled for all images in the batch (i.e. both instance and class) rather than only instance.