Training a new adversarial policy against kata1-b40c256-s11840935168-d2898845681 at 1600 playouts using the same hyperparameters we've been using thus far will fail to achieve a win rate of more than 50%.
Created by norabelrose 18 days ago; known in 9 days
- norabelrose estimated 70% 18 days ago
Please log in to respond to or judge prediction