Chess engine: Tucano 10 NNUE
Rating CEDR=3252
v.10.00
Neural Network Evaluation
-------------------------
Starting with version 10 release, Tucano uses a neural network evaluation, which increases the engine strength.
This network was trained on about 1.5 billion positions using Tucano's evaluation at depth 8 and Nodchip trainer's code (https://github.com/nodchip/Stockfish).
The neural network access code is from Daniel Shawn nnue library (https://github.com/dshawul/nnue-probe).
In order to use neural network you need to use an evaluation file, currently tucano_nn01.bin, that can be found in the Tucano’s github release section.
When running tucano, it will try to load the file from the same folder where tucano executable is.
You can load a different file by using a new command line parameter, e.g.: tucano -eval_file tucano_nn01.bin.
Also, you can send the file name through the new UCI option "name EvalFile type string".
It is important to make sure tucano can locate the eval file.
To validate the file is loaded you can start tucano on your system and see if it
Comments
Post a Comment