curious_maze

Example code for the paper "Intrinsic Rewards for Exploration without Harm from Observational Noise: A Simulation Study Based on the Free Energy Principle".

Train an agent without intrinsic rewards with

python run main.py

or adjust arguments like

python main.py --curiosity hidden_state --alpha None