First you need to connect to an available server. In addition to LeelaZero, you can try the KataGo server, but it seems a bit slower and like it has a bit more limited game tree search. All I know LeelaZero is based on the DeepMind's paper on AlphaZero, so it should be pretty strong. First you choose the game you want to review. You can upload it by using the menus on the game review screen that comes up first. The chart on the right hand side shows whether the game is biased to black (up, 0), or white (down, 1).

Today, Facebook AI Research (FAIR) open sourced ELF OpenGo, an AI bot that has defeated world champion professional Go players, based on our existing ELF platform for Reinforcement Learning Research. We are releasing both the trained model and the code used to create it. Inspired by DeepMind's work, we kicked off an effort earlier this year to reproduce their recent AlphaGoZero results using FAIR's Extensible, Lightweight Framework (ELF) for reinforcement learning research. The goal was to create an open source implementation of a system that would teach itself how to play Go at the level of a professional human player or better. By releasing our code and models we hoped to inspire others to think about new applications and research directions for this technology.