About the action selection in Double Q-Learning about reinforcement-learning-an-introduction HOT 4 CLOSED

Thanks for the contribution! That's true.
You may find this issue #37 interesting.
At the very beginning I also noticed this and implemented my own argmax and put it into a utils folder. Although I believe the usage is well documented, many people still complain about import error in the above issue. So about a month ago, I deleted my own argmax and use np.argmax all the time.

from reinforcement-learning-an-introduction.

@ShangtongZhang Thanks for your reply :D
By the way, I found the Example 6.7 Maximization Bias Example cannot reproduce book's plotting. Could I open a new issue?

from reinforcement-learning-an-introduction.

@ewanlee Sure. I'll look into that.

from reinforcement-learning-an-introduction.

@ShangtongZhang WOW! I found the plotting is true when I fixed this issue.
Thank you very much, anyway :P

from reinforcement-learning-an-introduction.