Implementing nother algorithm

[jaykodeveloper]

Hi Aleju,

It was really great work!! I want to implement my own calculation, but don't know where to start. If my code is reinforcement, can I just change code in train.py in train_reiforced directory?

Thanks,

[aleju]

Yes, I think all the reinforcement learning stuff is in train_reinforced/. The reward computation is in lib/rewards.py, possible actions in lib/actions.py, replay memory saving and loading in lib/replay_memory.py and the reading and communication with the game window are mainly in lib/ets2game.py and lib/ets2window.py.
Won't be a trivial task to change the algorithm. Scrolling through the files, it seems like I did not exactly aim for the highest code quality.