SakinaBohra/ppo-Huggy

PPO Agent Playing Huggy

This agent was trained using Unity ML-Agents as part of the Hugging Face Deep Reinforcement Learning Course.

Training

• Environment: Huggy
• Algorithm: PPO
• Framework: Unity ML-Agents 1.1.0
• Training steps: Approximately 2 million

Models

• Huggy.onnx: Final trained policy
• Huggy-1199973.onnx: Intermediate checkpoint

Training Results

• Highest recorded mean reward: 4.021 at 1.2M steps
• Final recorded mean reward: 3.633 at 2M steps

These are training rewards, not independent evaluation results.

Play

Open the Huggy browser simulator and select this model repository.