WellyZhang / mx-DDPG

MXNet Implementation of DDPG
0 stars 1 forks source link

target_update #2

Closed ShawnLue closed 7 years ago

ShawnLue commented 7 years ago

The update of target network is noneffective. I have verified this in my experiments, that the qfunc_loss is quickly move to 0 as a result of the static state of target network.