set train_scale_cfg=4.0 processed 3000 video, after train, the result still bad; similar to uncfg train;
Activity
01-111162-173 commentedon Sep 2, 2026
+1
zonszer commentedon Sep 2, 2026
@smith-fugui Hi, I am curious did you use fl2va or ref2va model for this?
smith-fugui commentedon Sep 3, 2026
fl2av