Dragon NNUE by Komodo Chess

 Dragon by Komodo Chess

Larry Kaufman:

We have released Dragon at komodochess.com. Dragon uses NNUE (Neural Network Updated Efficiently) technology, originally developed for the game of shogi. Komodo has a great deal of chess knowledge in its evaluation. Training an NNUE network based on this evaluation was both an advantage and a challenge, requiring experimentation with architectures and data generation of billions of positions. The reinforcement learning phase for Dragon is in its infancy, but is already showing great promise.

Dragon is a huge strength improvement over 14.1, the last release, about 197 elo in MCTS mode and 189 elo in standard mode, at CCRL blitz time control on one thread based on direct matches. With four threads, the gains were 156 elo in MCTS and 170 in standard mode. The improvement is due to the embedded neural network providing much more accurate evaluation and also in some sense gaining an extra ply or so of search by “seeing” some tactics that a normal eval won’t spot. Our testing also shows that Dragon is especially strong in Fischerandom (960) chess. We believe that Dragon will play in a more human-like style than standard Komodo since it relies on learning what actually works in games rather than just on pre-assigned values for eval terms.

There is one new feature added since Komodo 13.3, "Armageddon" mode, which tells Dragon that White (or Black) must win, draws are scored as losses for that color. When draws are counted as losses for White, knowing this improves Dragon’s performance as White by about 30 elo according to our tests. Other features added since Komodo 13 include more levels, Personalities, and Auto-Skill. Standard mode remains the strongest when Multi-PV is not in use, while MCTS mode is generally strongest when both are using Multi-PV. Skill levels have been adjusted to use the Dragon NNUE without much change in the strength of a given skill level; we think this makes the skill levels more "human-like".

More