Closed FarhaParveen919 closed 4 years ago
This repository doesn't consider to cover a wide range of RL environments so as to focus on delivering the basics of RL techniques as a beginner's tutorial. Unfortunately, I don't have any ipynb example for it, so you may need to employ things not introduced in this repository to run Atari environments. I feel sorry to say that.
Dear Jinwoo,
Thank you very much for putting up this code. Running it on Jupyter notebook has been very useful and informative.
I am trying to run the code (rainbow.ipynb) now on Atari games such as SpaceInvaders-v0.
I tried the following :
num_frames = 10000 memory_size = 1000000 batch_size = 32 target_update = 8000
train
agent = DQNAgent(env, memory_size, batch_size, target_update) agent.train(num_frames)
However, I am receiving a matrix mismatch error (attached). I realize that porting the code to Atari is not straight forward, and might need the techniques used in https://github.com/medipixel/rl_algorithms.
Before going through the above github in more detail, I was wondering whether you had an example of running any Atari game with the existing .ipynb notebook.
Regards, Farha