Closed Adrii98 closed 3 years ago
Although this technically is possible using cloneState / restoreState I'm not sure why you'd want to do this. Greedily choosing the action which maximizes the reward on the next time step isn't necessarily optimal. If you really want to do this try playing around with cloneState / restoreState, you should be able to do something like,
currentState = cloneState()
for every action:
restoreState(currentState)
execute action
observe reward
check which action resulted in the maximum reward
Hi, I have a question, i am trying to pass to the ale the action that gives the biggest reward in a certain moment, therefore i want to know how can i take the reward of a given action without using the function act. My goal i to know the biggest reward action and after that pass that action to the ALE act method. Thanks in advance