Increased training epochs from 1 to 2. This works with various SDXL1.0 based checkpoints but not good with pony.
This is an experimental lora model of an original AI character. based on CG and photos and GAN.
