The second sentence in the description: "Agents can be trained using reinforcement learning, imitation learning, neuroevolution, or other machine learning methods through a simple-to-use Python API." Are they lying?
OK, let me clarify: ML-Agents only provides a solution for reinforcement learning, and we will only help out with questions about reinforcement learning. If you want to use the low-level python API to train agents with any of those other things, you're welcome to, but you're on you own.