Chinese actress Zhao Jinmai's face mold Lora,the weight is between 0.7-1.2. I think about 0.8 is the most balanced. If you need to add some modifications to your face, please use 0.7-0.8. But in general, the weight of 1.0 is sufficient, and it can be very stable to generate the actress's face.This is an experimental LoRa that only trains the 7, 12, 16, and 20 floors of Block_dim.And I did not mark the picture. This theory is currently very popular, so I tried it. Because FLUX's parameters are large enough, it has seen too many descriptions, so unless it is a particularly unpopular concept, such a real -human LoRa training may be prompted.<h3 id="usage-disclaimer-tk68ng2z2">Usage Disclaimer</h3><ol><li>Disclaimer and Legal Compliance: Before using this LoRA model, users must carefully read and fully understand the associated disclaimer documentation. This model is intended solely for learning and research purposes and should not be used for any illegal or improper commercial purposes. Users are required to comply with all applicable laws and regulations in their respective jurisdictions when using the model.</li><li>Prohibition of Adult Content Generation: This LoRA model strictly prohibits the generation of any adult content. Users must adhere to this rule, and any violation may result in legal consequences.</li><li>Protection of Privacy and Image Rights: The images generated by this model must not infringe upon anyone's reputation, privacy, or image rights. Users must ensure that their use of the model does not infringe on the legal rights of third parties.</li><li>Non-Commercial Use Only: This LoRA model is based on the Flux 1-dev model and is subject to its non-commercial use license​(<a target="_blank" rel="ugc" href="https://fluxpro.art/posts/training-flux-lora">FluxPro</a>)​(<a target="_blank" rel="ugc" href="https://huggingface.co/XLabs-AI/flux-RealismLora">Hugging Face</a>).Therefore, users are prohibited from using the generated content for any commercial purposes.</li><li>Final Interpretation Rights: The developer retains the final interpretation rights of this model and reserves the right to modify the model and associated files as necessary.</li></ol>——————————————————中国女演员赵今麦的脸模lora，权重在0.7-1.2之间，我认为0.8左右是最平衡的，如果你需要在脸部加入一些修饰，请使用0.7-0.8。但一般而言，开1.0的权重，其泛化性也足够，而且能够非常稳定生成这位女演员的面部。这是一个实验性lora，仅对block_dim的7、12、16、20四层进行了训练。并且我没有对图片进行打标。这个理论目前非常流行，所以我尝试了一下。因为Flux的参数足够大，它已经见过了太多的描述，所以除非是特别冷门的概念，类似这种真人类型的lora训练，也许提示词打标不是必要的。<h3 id="-0yuy9rfy8">使用声明</h3><ol><li>免责声明与法律合规：在使用本LoRA模型之前，用户应仔细阅读并完全理解相关声明文件。模型仅供学习和研究目的使用，不得用于任何非法或不当的商业用途。用户必须在使用过程中遵守其所在区域的所有适用法律法规。</li><li>禁止成人内容生成：本LoRA模型严禁用于生成任何形式的成人内容。所有用户需严格遵守这一规定，否则将承担相应的法律责任。</li><li>隐私和肖像权保护：使用本模型生成的图像不得侵犯他人的名誉权、隐私权或肖像权。用户在使用此模型时，必须确保其行为不会对任何第三方的合法权益构成侵害。</li><li>非商业使用限制：本LoRA模型基于Flux 1-dev模型开发，遵循其非商业使用许可协议(<a target="_blank" rel="ugc" href="https://fluxpro.art/posts/training-flux-lora">FluxPro</a>)​(<a target="_blank" rel="ugc" href="https://huggingface.co/XLabs-AI/flux-RealismLora">Hugging Face</a>)。因此，用户不得将生成的内容用于任何形式的商业用途。</li><li>最终解释权：本模型的最终解释权归开发者所有。开发者保留根据需要对模型及相关文件进行修改的权利。</li></ol>

Zhao Jinmai 丨 赵今麦 丨 Chinese actress-Flux

权重：0.7-1.2，不需要触发词。

Files with SHA256: fa6c7f6677a3420fba937fc2e907f554f559855093cb6a5a1fd288496e1aad03