Peer Review Lab 3 - Leonor Gomes

[leonormgomes]

• The README is really clear and appreciated
• As a suggestion for the reinforcement learning agent, you could make it learn incrementally by the quality of the other agents. As for an example, first you could make it play against the "dumb" agent, then move to random and then move to the optimal player. You could even create new agents that are between dumb - random and random - optimal to have a smoother learning curve
• The coding coverage is reasonable, I think some comments from previous lines of code could be deleted to make it cleaner. It was probably by accident, but I also found a comment in italian, I would say to keep the comments in english.
• Regarding the evolved agent strategy, have you considered reducing the "GENETIC_OPERATOR_RANDOMNESS" value over time so that you can play a bit more with Exploration VS Exploitation?

[GabriG23]

Thank you Leonor for all the reviews, sorry for the response time, it's been a busy period. For the evolved agent if I remember correcty, yes we have reduced the genetic_operator_randomness value but it was not mutating as we wanted to so we kept it high. Good luck with exams!