Multiplayer AlphaZero

Training AlphaZero for 700,000 steps. Elo ratings were computed from

Combining Deep Reinforcement Learning and Search for Imperfect-Information Games

Multiplayer AlphaZero – arXiv Vanity

Alpha Zero / MuZero differences · Issue #143 · werner-duvaud/muzero-general · GitHub

Adversarial Search

AlphaDDA: strategies for adjusting the playing strength of a fully trained AlphaZero system to a suitable human training partner [PeerJ]

Deep learning for general game playing with Ludii and Polygames - IOS Press

DeepMind reveals more about MuZero, its A.I. that masters games without knowing the rules

Superhuman AI Triumphs Playing the Toughest Board Games

uttt.ai: AlphaZero-like AI self-play for Ultimate Tic-Tac-Toe with 100,000 simulations per move