Introducing Snowball Fight ☃️, our first ML-Agents environment

Hugging Face Blog · 4 years ago

Hugging Face released Snowball Fight, a custom Deep Reinforcement Learning environment built with Unity ML-Agents where players compete against AI agents, playable online through Hugging Face Spaces. The environment is open source and hosted on Hugging Face Hub, with plans for a 2vs2 cooperative version using the MA-POCA algorithm in coming weeks. This launch establishes an ecosystem for ML-Agents researchers to build, share, and host custom environments and trained models.