iteration datasets

AlphaZero Training Run: 62-Iteration Self-Play Metrics

Policy loss, value loss, game length, and MCTS agreement metrics across 62 iterations of AlphaZero-style reinforcement learning.