Open veenan2791 opened 2 years ago
There are two thing you might need to check:
Yes, it could be the random initialization issue. I ran the code multiple times before. Although, it worked after that. It works without error once in at least 5 runs. I will also look at the hyperparameter tuning in the model architecture. Thank you very much for your suggestions.
I am using the Bezier GAN model on my own dataset with different closed shapes. However, it stops training at random training steps and produces nan fake data throwing an assertion error. The error occurs from the below code line in the beziergan/gan.py file.
Here is the kind of error I get.
Does anyone know the reason behind it and how to resolve the issue? It would be a great help for me.