-
Notifications
You must be signed in to change notification settings - Fork 126
Trainer Controller
VinF edited this page Mar 21, 2016
·
2 revisions
A controller that makes the agent train on its current database periodically.
- evaluateOn [str] - After what type of event the agent shoud be trained periodically. Possible values: 'action', 'episode', 'epoch'. The first training will occur after the first occurence of [evaluateOn].
- periodicity [int] - How many [evaluateOn] are necessary before a training occurs
- showAvgBellmanResidual [bool] - Whether to show an informative message after each episode end (and after a training if [evaluateOn] is 'episode') about the average bellman residual of this episode
- showEpisodeAvgVValue [bool] - Whether to show an informative message after each episode end (and after a training if [evaluateOn] is 'episode') about the average V value of this episode