Closed adizhol closed 5 years ago
Dear Adi,
thanks for the question! This is done in the code to confuse the reader.... Kidding! :-) The thing is this is a legacy from the older code, where we did some hyperparam tuning and there was no easy way to merge the code, so we ended up hard-coding these constants. They only affect the lambda / learning_rate.
Okay, thank you :)
Hello Sir,
I saw that in the reconstruction_loss function (wae.py) you multiply L2, L2_2 and L1 losses with different kind of values, but I haven't read anything about this in the paper. Can you please explain?
Thank you, Adi Zholkover