Published On Aug 2, 2015
When the Google DQN was trained to around 5 million steps, it seems that the learning is converge, but I hope it will continue to learn more and better, this video is the screen cast of the simple analysis I did.
show more
When the Google DQN was trained to around 5 million steps, it seems that the learning is converge, but I hope it will continue to learn more and better, this video is the screen cast of the simple analysis I did.
