lora training with my deliciousangerous model
I find that tags work better than complex prompts
babel-dark.safetensors