基于SDXL1.0训练的LoRA模型，权重控制在0.8以下可以产出非常帅气的男性人像
推荐使用DPM++ 2M SDE Karras采样器，并且迭代步数设定为20以达到最佳质量
The LoRA model trained on SDXL1.0, with weights controlled below 0.8, can produce very handsome male portraits
It is recommended to use a DPM++2M SDE Karras sampler with an iteration step of 20 to achieve optimal quality