Open hamiltonwei opened 6 years ago
Did you make any progress on this or find an implementation that allowed you to do this?
Sorry I have stopped working on this project for a while so my memory is getting rusty. One progress I did make is that using the breakout RAM as the input doesn't seem to work. I have to use the frames as the input to make the agent work. I am not sure why that's the case though as I haven't investigated this problem too deeply.
I am trying to adapt your code to train the agent to play breakout. I tried to use both the CartPole-basic file as well as the Seaquest-DDQN-PER file but the agent doesn't seem to learn after training for couple of hundreds of episodes (the total reward is around 1 to 3 in average). Have you tried to train the agent to play breakout with the code? If so was rgere any effects? I was using "Breakout-ram-v0" for the CartPole-basic file and "Breakout-v0" for the Seaquest-DDQN-PER file.