I want to introduce you to Annie, a chess bot for Lichess. She's excited about exploring the more intricate sides of chess. Her favorite openings are the cloud variations and she is a very enthusiastic fan of en passant in every imaginable form.
Annie's handcrafted, large-table powered evaluation was trained on no fewer than six million, four hundred and thirty-four positions from games played on Lichess. And not just the games of grandmasters, but the games of noobs and sub-800 Elo players, too. On top of that, she also has an anarchic sub-module in her tree search. So be prepared for some unconventional moves (and comments).
You can play Annie here (bullet, blitz, or rapid, always with increment and unrated).
Depending on how well you've played recently, a level from A1 to C3 will be suggested. If you only understand the most established strategies that grandmasters tend to go for, you better play against Annie at level C1, C2, or even C3. But if you have experience battling the chaos of playing against 800 Elo prodigies and if you can keep your composure when facing an opponent with a question mark next to their rating, you should play against level A3, A2, or if you're really mad, against level A1. All you players who are ordinary that it hurts, give the B levels a try.
Annie is based on the chess engine Nalwald.
Copy config.default.json to config.json and replace the place-holder for "lichessToken" with your lichess bot account token. You can also edit the other available config parameters.
Now run ./runLichessBot.sh and you can start playing the bot on lichess!
(Some of the code also has an anarchy feeling to it, but that's definitely intended ...)
You need the Nim compiler (version 1.6 or higher) and the Clang compiler.
nim default Annie.nim
The resulting binary can be used as a UCI chess engine.
There will be an UCI setting called DifficultyLevel which allows to select the strength of the engine, 1 being the weakest and wildest, and 10 being the strongest (but also most boring) setting.
Pre-compiled binaries can be found here.
Copyright © Jost Triller