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.

Sign up for GitHub

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

Jump to bottom

Prompt used during pretraining #17

Open
Actis92 opened this issue Apr 6, 2023 · 1 comment
Open

Prompt used during pretraining #17

Actis92 opened this issue Apr 6, 2023 · 1 comment

Comments

@Actis92
Copy link

Actis92 commented Apr 6, 2023

Why you have decided to use empty prompt "" when you need to use the unet to build the features from the noise image?
input_ids_for_encoder = tokenizer( "", # args.prompt_template.format(placeholder_token=args.domain_class_token), padding="max_length", truncation=True, max_length=tokenizer.model_max_length, return_tensors="pt" ).input_ids
We know that the image would be something like "a photo of args.domain_class_token" so I'm not sure if this can have an impact on the pretraining
@mkshing
Copy link
Owner

mkshing commented May 22, 2023 •
edited
Loading

@Actis92 Hi sorry for the late reply. This is actually a good point! You can definitely experiment this. In fact, you can see this line to show that I was about to try it. But, I ended up to use the null string because the author apparently used it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@mkshing @Actis92