Experiment 25

December 1, 2025 23 minute read

This post demonstrates that reinforcement learning agents can be trained with sparse reward signals as effectively as carefully tuned dense rewards.