Skip to content

Configuration for FFHQ dataset #20

@cxw-droid

Description

@cxw-droid

Hi,

We want to train the consistency models on FFHQ dataset (64x64). Could you please give some suggestions on how to set the parameters for this dataset? I tried to set the parameters following this, and found the best FID is roughly 4, much worse than claimed value in the EDM paper. It's expected to get a similar or better FID value with CM training.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions