It's not running correctly.

It keeps giving me:

AttributeError: module 'tensorflow' has no attribute 'reset_default_graph'

whenever I try to run it. How do I fix this?

[Kirozen2020]

did you fix that? , I have same problem

[DirckM]

It keeps giving me:

AttributeError: module 'tensorflow' has no attribute 'reset_default_graph'

whenever I try to run it. How do I fix this?

Have you found a solution for the problem because I am running into it as well?

[ArtsyMedia]

It keeps giving me:
AttributeError: module 'tensorflow' has no attribute 'reset_default_graph'
whenever I try to run it. How do I fix this?

Have you found a solution for the problem because I am running into it as well?

nope. Not yet.

[breadbrowser]

you need older python version

[AlexandreCGMeier]

I have another issue now. main.py, Line 195 "QValueOfNextStates = self.sess.run(self.TargetNetwork.output,
feed_dict={self.TargetNetwork.inputs_: nextStatesFromBatch})"
I get the error: "Cannot feed value of shape (64,) for Tensor TargetNetwork/inputs:0, which has shape (None, 15)". I've found a StackOverflow Discussion which seems to be relevant, but I don't see how to apply it https://stackoverflow.com/questions/37433321/tensorflow-tflearn-valueerror-cannot-feed-value-of-shape-64-for-tensor-uta

EDIT: Nevermind, it was weird bug that I didn't understand but "modernizing" QLearningfromMate.py solved the problem (make it compatible with TF2). Feel free to clone my repo as it is functional. I also included optimization to run on CPU.

[Pommes203040]

its 2023 and its not working but i wourld try it so much :(
because its looks very cool pls fix it pls

[LjAquinox]

@Pommes203040 I spent the last 3 days fixing things and making a more user friendly version I'll upload it today if you are still interrested try it for yourself

[ArtsyMedia]

@Pommes203040 I spent the last 3 days fixing things and making a more user friendly version I'll upload it today if you are still interrested try it for yourself

yes please!

[LjAquinox]

@ArtsyMedia well I just created it. Link : https://github.com/LjAquinox/Car-QLearning

[AlexandreCGMeier]

If anyone is still interested. I just finished the final debug and after training for 3h on my CPU (approx. 6k episodes) your AI gets to clear the course. Check out my repo. Train_QLearning to train, and then Render_QLearning (line 56: starting_episode = YOUR NEWEST CHECKPOINT HERE) shows the newest AI. Next I'll rewrite from scratch in PyTorch as an exercise. Does anyone have a reference implementation I could refer to? I know PyTorch but never programmd DDQN in it. Thanks.

[LjAquinox]

If anyone is still interested. I just finished the final debug and after training for 3h on my CPU (approx. 6k episodes) your AI gets to clear the course. Check out my repo. Train_QLearning to train, and then Render_QLearning (line 56: starting_episode = YOUR NEWEST CHECKPOINT HERE) shows the newest AI. Next I'll rewrite from scratch in PyTorch as an exercise. Does anyone have a reference implementation I could refer to? I know PyTorch but never programmd DDQN in it. Thanks.

I like the fact that this code is still being used 😄 I think there is a lot of added complexity in your current implementation and I kind of don't have enough time to read trought it all to make comments but if you continue to work on this I'll take a look at it another time.

I have not implemented Q networks in pytorch either so I wont be able to help you with that either.
Good luck 😅