Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
JunJun-Liu 44e20808c7 | 7 months ago | |
---|---|---|
ScienceChina-data | 7 months ago | |
configs | 8 months ago | |
demo | 9 months ago | |
ldm | 7 months ago | |
mindpet | 8 months ago | |
readme | 8 months ago | |
scripts | 8 months ago | |
README.md | 8 months ago | |
README_EN.md | 9 months ago | |
inference-qizhi.py | 8 months ago | |
inpaint.py | 9 months ago | |
openi.py | 9 months ago | |
requirements.txt | 9 months ago | |
run_db_train.py | 9 months ago | |
run_train.py | 9 months ago | |
train-qizhi-lora.py | 8 months ago | |
train-qizhi.py | 8 months ago | |
train-zhisuan.py | 7 months ago | |
train.py | 9 months ago | |
txt2img-zhisuan.py | 7 months ago | |
txt2img.py | 7 months ago |
首先基于原始悟空画画代码微调仓库,作为下述所有复现步骤的基础。
链接:https://github.com/mindspore-lab/minddiffusion/tree/main/vision/wukong-huahua
fork本仓库,并创建一个调试任务,环境选择如下:
新建之后,上传本仓库的代码文件。
上传完成后,解压zip数据集的包:
我们按照个性化微调的步骤,复现自定义的主题(比赛主题):科幻中国
执行bash scripts/run_db_train.sh,训练个性化微调后的自定义模型权重文件,效果如下:
如果训练正常,log_tain文件应该如下所示:
悟空画画微调
Python Text Markdown Shell Unity3D Asset
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》