Open Adubi opened 1 year ago
@Adubi Hello, I was having the same problem and found a solution in pull requests in this project. You have to change the initial value of mx_nxt_reward in chooseAction function. Check out the pull requests or this link to see full detail.
https://github.com/MJeremy2017/reinforcement-learning-implementation/pull/11
i found a bug in the q-learning code...... it runs but sometimes while running just stops and gives a key error