-
Notifications
You must be signed in to change notification settings - Fork 21
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add image as control condition #21
Comments
Hi! Could you share a picture with the training loss? I assume that by “starts oscilating and does not decrease” you mean the loss right? |
Hi, the loss for diffusion is in general quite unstable, usually quickly converges and doesn’t change much (see this issue). So I would suggest you to the model training for a few epochs and check the results during the intermediary checkpoints. If the results are not good you can share it here or via email with me and I will be more than happy to discuss it and try helping you make it work conditioned to images as well. :) |
Hey, I wonder that how did you add images as control condition in this task? Could you briefly introduce your methods? |
I tried adding images as more control conditions, but our model starts oscillating after the second epoch and does not decrease. Has the author done this experiment before?
The text was updated successfully, but these errors were encountered: