[SD3.5] 3DKitten

Update v1.1 Freeze 30~37 layers for lora training.

a simple testing lora for sd3.5

Trained by T2ITrainer: https://github.com/lrzjason/T2ITrainer

100 images

4 rank

5 repeats

1 batch size

2 gradient_accumulation_steps

prodigy