Trained on small dataset what i gather from internet. To use simple use regina in tags and <lora:ReginaDinoCrisis:0.9>. For better result use around 0.7-0.9 weights of lora. Trained on deliberate_v2 model.