Tag
GoBench is a benchmark for evaluating large language models on 9x9 Go games, demonstrating strong correlation with ARC-AGI and featuring a leaderboard with KataGo opponents from random to superhuman levels.
Go grandmaster Shin Jin-seo defeated AI engine KataGo in a historic three-match series, marking the first human victory against a state-of-the-art Go AI under a two-stone handicap.
A study on how KataGo's Go-playing neural networks internally represent board symmetries, finding that the model learns partially symmetric concepts despite only stochastic 8-fold data augmentation, with one unexpected result.
Tests a zero-parameter model against KataGo, evaluating its performance in the game of Go.