The implementation described in the paper is different from the actual implementation of the shared code. For example, the weights of the loss functions are different, and there are differences in the use of squared or absolute values.
What should I follow to get the exact implementation?
The implementation described in the paper is different from the actual implementation of the shared code. For example, the weights of the loss functions are different, and there are differences in the use of squared or absolute values.
What should I follow to get the exact implementation?
Thanks!