An agent that seeks +1, avoids −1. Set the reward on the right →, then add demonstrations and watch it move.

MaxEnt-style inverse RL: reward ∝ where the agent is drawn (green) vs. avoids (red), vs. a random walker. (The repo's GenJAX backbones do the Bayesian-inversion version.)

The reward you set — click any cell here to cycle +1 → 0 → −1.