I have trained reinforcement learning model with windowed time-series data and now i am getting inference. But i have a doubt that model(batch) works in order to Last IN First OUT. Cause, otherwise it seems like the last action is the first which always leave the action_list.
(An action is being decided in respect to the current one line and past 19 line of data.)
Is there anybody who knows which order the model function works in Pytorch, First in First out or Last in First Out?