This was trained using the CivitAI LoRA trainer on several of my Krea 2 images generated using the VXP 3.0 and 3.1 models.
Disclaimer
Responsibility for generated content falls on the user of this lora. I do not control how people prompt or decide what their intent is.
Description
This version was trained with MOSS Base instead of MOSS SFT.
The dataset is exactly the same. I've seen a few places saying training with the base model is recommended, so here we are.
I use it at 0.70 to 0.80 strength.