You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am curious why are the weights used in the critic and actor networks being initialised using orthogonal matrices. What is the motivation of doing this approach instead of xavier initialisation?
Hi,
I am curious why are the weights used in the critic and actor networks being initialised using orthogonal matrices. What is the motivation of doing this approach instead of xavier initialisation?
url_benchmark/utils.py
Line 62 in bb98f0c
The text was updated successfully, but these errors were encountered: