For SDXL arch, it's a LoRA model USING the LoHa algorithm in LyCORIS

For Anima arch, its a NORMAL LoRA

bs10, 420 images, training was completed on a NVIDIA GeForce RTX 4090 GPU.

CLIP skip is not applicable to all SDXL architecture models, so there is no need to set it.

Trigger words:

thalassa

or use it with Anima model:

@thalassa

also, there's a secondary trigger word for nothing:

caspian

Reproduction without permission is prohibited.