Hey! This is a great repo, thank you so much to make available this information here. Unfortunately, a few libraries versions that you were using at the moment are obsolete. Therefore, the behavior when running the notebooks are completely different from what is expected. In my case, I tried the DDQN notebook, but the training is not steady and it doesn't converge (never reaches the goal reward of 195). Is it possible to update the notebooks? I think that would be a great contribution.
Thanks!
Hey! This is a great repo, thank you so much to make available this information here. Unfortunately, a few libraries versions that you were using at the moment are obsolete. Therefore, the behavior when running the notebooks are completely different from what is expected. In my case, I tried the DDQN notebook, but the training is not steady and it doesn't converge (never reaches the goal reward of 195). Is it possible to update the notebooks? I think that would be a great contribution.
Thanks!