train_scale_cfg · Issue #1663 · modelscope/DiffSynth-Studio · GitHub
Skip to content

train_scale_cfg #1663

Open
Open

Description

@smith-fugui

set train_scale_cfg=4.0 processed 3000 video, after train, the result still bad; similar to uncfg train;

Activity

  1. 01-111162-173 commented on Sep 2, 2026

    @01-111162-173

    +1

  2. zonszer commented on Sep 2, 2026

    @zonszer

    @smith-fugui Hi, I am curious did you use fl2va or ref2va model for this?

  3. smith-fugui commented on Sep 3, 2026

    @smith-fugui
    Author

    @smith-fugui Hi, I am curious did you use fl2va or ref2va model for this?

    fl2av

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions