Reinforcement learning

reward sparsity

Sparse success increases credit-assignment variance and can over-reward lucky trajectories.

1 source videos