xsmxsm

@wmonica yes, I have soveled it. Because there are some errors in my training model, so it cant't infer success. the error is lhs shape= [40,1,1,128] rhs shape= [1,1,1,128] in...

@wmonica you have to change lhs shape=rhs shape,then the error can sovel. maybe you should change your embedding_num=40(or other number), same in train.py and unet.py you should try it, because...

@sellinaxmj 你的环境没有配置好，缺少imageio模块，先安装好imageio再训练

@whyseu 你需要在工作的文件夹下新建一个文件夹，比如文件为text，然后让--sample_dir=text