Marathon Environment from Unity ML-Agents

Training the Unity ml-agent's marathon environment.

Unity ML-Agents’ marathon environment contains a lot of commonly used environments that are similar to ones used in modern AI research. It is easy to use our in-editor training to train the agents in the marathon environment as well.

Trained Deepmind Hopper

Go to Sourcecode

You need the UnityTensorflowKeras repository to train ML-Agents in editor. Go to the repository from the link below to install it according to the instructions.

The marathon environments are not included in ML-Agents or UnityTensorflowKeras repositories. You need to download it from its own repository and put the assets into your Unity project.

Exercises

1. Pick one of the environments from the marathon environments, setup the scene for training in editor, and train it. You can use the same parameters for neural network and training as specified in config/marathon_envs_config.yaml file.