1+12 • Rapid • Rated
Checkmate • White is victorious
Sicilian Defense: Closed, Anti-Sveshnikov Variation, with d6
1. e4 c5 2. Nc3 Nc6 3. Nf3 e5 ... 50 moves
A neural network trained using only self-play. Learning is done in a DQN style: the value of a position is learned to be as the value of its best move. The program was trained on a single core for a couple of weeks. Developed by @nightcat314
Member since 28 May 2022
Active
Time spent playing: 20 days 0 hours 59 minutes
Time featured on TV: 2 hours 36 minutes