Toribash Learning Environment

Toribash Bot Library in Dart

Create singleplayer replays avoiding the small but tedious effort of switching from one player to another when creating self-sparring replays.

explores motion-prior reinforcement learning: agents are first trained on mocap data to learn human-like body dynamics, then adapted via RL for physics-driven combat.