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

Discrete diffusion in the code #3

Open
vadimkantorov opened this issue Jul 27, 2023 · 1 comment
Open

Discrete diffusion in the code #3

vadimkantorov opened this issue Jul 27, 2023 · 1 comment

Comments

@vadimkantorov
Copy link

Hi! Thanks a lot for releasing the codes :)

Do I understand correctly that the main model is in https://github.com/OpenGVLab/DDPS/blob/main/mmseg_custom/models/decode_heads/segformer_head_unet_fc_head_single_step.py#L128 ?

Am I correct that first the discrete class labels are embedded into real-valued vectors, put into the unet which outputs real-valued logits? (and that corruption is done in the discrete label space)

@Siddharth-Shrivastava7
Copy link

Siddharth-Shrivastava7 commented Jul 31, 2023

Hi @vadimkantorov ,

As far as I know, this codebase is similar to implementation of "Argmax Flows and Multinomial Diffusion:
Learning Categorical Distributions", Reference: https://github.com/ehoogeboom/multinomial_diffusion/blob/9d907a60536ad793efd6d2a6067b3c3d6ba9fce7/segmentation_diffusion/layers/layers.py#L190

Hope this helps!

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