This model works better with a 1.3 strengh, I also recommend using the "anime" tag for best results and to reinforce the style,
lora-000002.TA_trained (1).safetensors