/mario-rl

Super Mario Reinforcement Learning (RL) Training Colab notebook with Stable Baseline3 Library

Primary LanguageJupyter Notebook

Welcome!

This repo include a Super Mario Reinforcement Learning (RL) Training Colab notebook with Stable Baseline3 Library.

This is 1h slide deck for my colleagues and 1.5h slide deck for junior students.

Training progress review (note best 1 out of 20 applied to "cheat")

Steps 200,000 (fail) 500,000 (success) 1,000,000 (fail) 1,500,000 (success)
Best 1 out of 20 200k 500k 1m 1.5m