Skip to content
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

predicted x_0 #8

Open
betterze opened this issue Jun 3, 2023 · 2 comments
Open

predicted x_0 #8

betterze opened this issue Jun 3, 2023 · 2 comments

Comments

@betterze
Copy link

betterze commented Jun 3, 2023

Dear Zhao,

Thank you for sharing this great work, I really like it.

I am not an expert in sampling. In ddim, we can get the predicted x_0 in each sampling step.

image

Could we get the predicted x_0 in each sampling step from Unipc?

Thank you for your help.

Best Wishes,

Zongze

@wl-zhao
Copy link
Owner

wl-zhao commented Jun 5, 2023

Hi,

Thanks for your interest in our work! The $x_0$ can be computed just as in DDIM. You can refer to this function for details.

Best,
Wenliang.

@betterze
Copy link
Author

betterze commented Jun 6, 2023

I try to visualize the dynamic of ddim and unipc, and plot the x_t and predicted x_0 over time. I use 50 sampling steps. It seems like the predicted x_0 of ddim is better than that of unipc in early time steps, could you tell me why this happens? thx

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants