Curiosity-Critic: Cumulative Prediction Error Improvement as a Tractable Intrinsic Reward for World Model Training

Implement a Curiosity-Critic module that can be plugged into existing RL agents to improve exploration efficiency. Developers building autonomous agents will find this significantly boosts training convergence in sparse-reward environments.

Suggested repo: curio-critic

"Stop guessing where to explore; use cumulative prediction improvement as a reward."

Estimated effort: 90h