Supplementary Materials Evaluation beyond Task Performance: Analyzing Concepts in AlphaZero playing Hex

Appendix A reports implementation details, hyperparameters and compute requirements. Appendix B gives more details on each concept introduced in the main body of the paper. Appendix C demonstrates how AlphaZero often wastes moves. Appendix D has additional results across the different architectures. We use agents trained by Jones [5]. See Table 1 for hyperparameters and relative agent strengths.