jonathan-laurent / AlphaZero.jl

A generic, simple and fast implementation of Deepmind's AlphaZero algorithm.

https://jonathan-laurent.github.io/AlphaZero.jl/stable/

How important is loss during learning

StepHaze opened this issue a year ago · comments

StepHaze commented a year ago

If loss almost isn't changing (~1.06) during learning, but the network is replacing itself, is it normal?