Is the R L Agent Learning How to Solve a Task?

In the generalization trials, the r l agent takes longer to complete the task than it does in the trading trialsthing. But because the model is recurrent, now, you can think of that as like having extra time to sort of due some integration of information. And i can't say for sure what's happening, but that's one possibility for what it's doing. Itits sort of like integrating information time and getting more sure of its decision.