Final Report: Here Website: Here
Update: I migrate everything previously here to the timeline.md. Post what you did and how we can use your code here.
May 31
-
Delete preped data and parsed replay because we hit github storage upper limit. Download here in Google drive .
-
Upload pytorch_dqn_agent. Run it ./run_agent.sh. I am still fixing bugs, but it is pretty promising. I tried the author's other two agents and both worked.
To-do: Add the refined agent's actions to the DQN agent.