cookbenjamin / ddpg Goto Github PK

Hi Cookbenjamin, can you add the code for a sample world. In the main.py there is a world class but I was unable to find its code.

I must say your implementation of DDPG is really great. I was trying to code DDPG directly form the paper but I had some ambiguities but your code really saved me. It would be great if you can even point out how should I write a ''World'' class.

Thank you.

Hello! It appears there are some significant gaps in the provided code. There are various instance variables missing from the initialization methods. While the missing code can generally be inferred, the following code on line 127 of ddpg.py presents a problem. The _critic object does not have a method train. What is intended? Or, what is the train method for the critic?

self._critic.train(states, actions, q_targets)

Hi @cookbenjamin , could you please add the code for Critic? I only find codes for CriticNetwork..

Thanks!
Marx