Closed deahhh closed 2 years ago
In the paper, wasserstein distance is estimated with a new network f by minimum the minus of the two distribution, which is not displayed in this repository.
I don't understand your explanation, meaning that the code doesn't implement EMD?
I can see what you mean by reading this blog.
In the paper, wasserstein distance is estimated with a new network f by minimum the minus of the two distribution, which is not displayed in this repository.