I am getting an error when applying the code in any of the continuous action problems. The error I am getting is: Cannot feed value of shape (1, 3, 1) for Tensor 'Placeholder:0', which has shape '(?, 3)'. Any suggestion?
Great video thank you very much! Do you think you can make a tutorial on how to make reinforcement learning on non-gym environment? for example, an agent that can learn to play Agar.io only from looking at the screen and exploring the env. Again, many thanks for the awesome videos you put on this channel :)
this is the most elegant DRL tutorial on youtube thank you so much !!
thank you very much for creating this channel and sharing your knowledge!
Great video please add Dueling DQN Algorithm for CartPole
Thanks for your service
Great videos! Will you cover other models like policy gradients?
I am getting an error when applying the code in any of the continuous action problems. The error I am getting is: Cannot feed value of shape (1, 3, 1) for Tensor 'Placeholder:0', which has shape '(?, 3)'.
Any suggestion?
same problem here. Wonder if it's tf version problem?
Same
Change this in DiscretizedDQNAgent.get_env_action():
if not self.is_discrete:
action = [self.actions[action, 0]]
return action
Great video thank you very much!
Do you think you can make a tutorial on how to make reinforcement learning on non-gym environment?
for example, an agent that can learn to play Agar.io only from looking at the screen and exploring the env.
Again, many thanks for the awesome videos you put on this channel :)
Yes I agree! I'm working on some custom RL environments for my next video and hope to progress toward something like what you mentioned