jiqizaisikao

Hi,i have also met this problem

If we train the model on GPU ,could it predict on cpu?

Hi,Im using some Chinese database to train too,which data do you use to train net1 Contact Q:3617 28654

  这是我自己模型训练的结果，类似于Adobe的生成填充，微信最新的小程序，根据一张照片，一个模版，进行一键换头，一键换脸型。 不过需要不小的算力进行训练，我可能没有那么多资源进行训练优化了，希望可以找到朋友进行合作研究和开发



+1

@zhf459 I think is that when you use logistic mixture model,at the first flow end you sample wave like result as the input of the next input ,and so on...

Ok,i will try again

@zhf459 i have test power_loss and it works right,but i do not know how to complete the crossentropy loss,have you try it?