Create
Save
Share
Belpheg0r_
The training dataset was cleaned up, and the LoRA rank reduced from 64 to 32