CARA is a full-featured desktop application designed for analyzing and reviewing chess games. Unlike chess playing applications or puzzle solvers, CARA focuses exclusively on post-game analysis and ...
Combining supervised pretraining on master-level games with AlphaZero-style self-play reinforcement learning, using a SE-ResNet architecture with spatial attention and Win/Draw/Loss value estimation.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results