Ray00007/mlagents-SoccerTwos-POCA-AIVS

ML-Agents POCA model for SoccerTwos

This is a model trained using POCA (Proximal Policy Optimization with Centralized Actor) for the SoccerTwos environment from Unity ML-Agents.

• Run ID: Soccersend
• Training ELO: 1532.72 (at ~5M steps)
• Config: See SoccerTwos.yaml for all hyperparameters.

How to use (with ML-Agents):

1. Download the SoccerTwos.onnx and SoccerTwos.yaml files.
2. Create a folder structure results/Soccersend/
3. Place SoccerTwos.onnx inside results/Soccersend/
4. Download the SoccerTwos environment.
5. Run the inference command:

mlagents-learn ./SoccerTwos.yaml --env= --run-id=Soccersend --inference --resume