Andrej Karpathy - Missteps in Reinforcement Learning (via searchlore.ai)