The function of Encoder and the MMD code

[Ryann-Ran]

Hi there,
Thank you for your valuable contribution to the project.

I've been going through the code for training and visualization, and I have a couple of questions regarding the functionality of the Encoders.
In train.py, during the Encoder training phase, I noticed that the parameters of D and G are not updated. Could you please clarify how the Encoders function in the training process, especially in relation to the G and D?
Similarly, in visualization.py, the Encoder is loaded, but it appears not to impact image generation. Can you explain the intended role of the Encoder in this context?

Additionally, I encountered an issue while examining the line test_statistics, ret_matrix = mmd(sample_1, sample_2, alphas='median', ret_matrix=True) within the calculate_mmd function in fid_score.py. It seems that the parameter alphas is expected to be of float type, yet it is passed as a string 'median'. This has caused difficulties in running the MMD calculation successfully. Could you provide some insights on how to resolve this issue?

I would greatly appreciate your assistance in clarifying these aspects of the code. Thank you in advance for your help!