I have an improved version of Rhetoric. The tunning algorithm has been changed and corrected, and now the evaluation is quite different.
In fact, the similarity test between the different versions of rhetoric shows something very curious:
Code: Select all
sim version 3
------ Rhetoric 1.4.1 (time: 50 ms scale: 1.0) ------
95.33 Rhetoric 1.4.1 (bis) (time: 50 ms scale: 1.0)
53.87 Rhetoric 1.2 x64 (time: 50 ms scale: 1.0)
47.27 Rhetoric 1.4 x64 (time: 50 ms scale: 1.0)
44.63 Fruit 2.1 (time: 50 ms scale: 1.0)
43.41 RobboLito 0.21Q x64 (time: 50 ms scale: 1.0)
Changes:
- Tunning/learning changed. Using self play with time odds.
- Added UCI_LimitStrength and UCI_Elo options. Very simple, adding a delay in eval.
- Added an option to change the material weigth. You can change the style to make it more positional or more materialistic. Changing the material value from 50% to 200%. The very first game that I played using 50% of the material, was a crushing Tal-style game. Really entertaining, believe me.
- Added an evaluation cache.
http://www.chessrhetoric.com
Well, I would like to see a lot of downloads.
Click on "Descarga ya!"
Regards.
Alberto.