-
-
Notifications
You must be signed in to change notification settings - Fork 753
Open
Description
1.I used 3000 pairs of face datasets and anime face datasets for training. train_photo stores 3000 pieces of real face data, and anime stores two folders of style and smooth, each with 3000 anime face datasets, val, Each test stores 100 datasets of real faces. Is there a problem with the settings?
2.The above problem is estimated to be related to tensorflow's cuda. Finally, the environment was reconfigured and the training was normal. However, the result after training did not change, and the animation of the face could not be seen. Is it a problem with the training epoch, or a problem with the data set setting?
Metadata
Metadata
Assignees
Labels
No labels