14.2
1. New and stronger neural network made from 1.6 billion position analyzed with Rebel 14.1, estimated elo gain 50-60.
2. Search changes, estimated elo gain 15-20.
3. NNUE tuner parameter in Engine Settings. Default setting now 40%.
Self playing against 14.1 shows an elo increase of ~80-85.
On the Gambit Rating List Rebel 14.2 is rated 3278 at place 15.
Download Rebel 14.2
Download Source code
14.2
___________________________________________________________________________________________________
14.1
Changes
1. New and bigger neural network made from 1.2 billion position analyzed with Benjamin 1.1, estimated elo gain ~60.
2. Search changes, estimated elo gain ~40-50.
3. NNUE tuner parameter in Engine Settings. Changing the default of 100% to 125 will make Rebel even more aggressive, 150 up to the point of reckless, values lower than 100 will change the playing style to be more moderate. We tested the value 75 and it gave an extra 10 elo but it was preferred to leave the value at 100 exactly what the learner had created to keep the original Benjamin 1.1 playing style.
Credits
____________________________________________________________________________________________________
This year (2022) it's 42 years ago I started chess programming, from REBEL 1.0 -> REBEL 12 and after my retirement from the competition and commerce (2003) the free hobby versions of ProDeo only gained about 200-250 elo through the years 2003-2021.
Inspiration (after 19 years) returned after a well known chess programmer offered his help with the implementation of the NNUE technique.
And the result speaks for itself, an improvement of 300-350 elo. It's my pleasure to realese REBEL 14, back to my brand name and its roots.
Rebel 14 is released under the GPLv3 licence.
Technical
Rebel 14 consists of 4 parts,
1. The famous Fruit 2.1 engine from 2004 by Fabien Letouzey, elo ~2700
2. Growing Fruit by Pawel Koziol improving the Fruit search, elo ~2800 making it about equal in strength with ProDeo 3.1
3. Replacing the Fruit evaluation with Benjamin 1.1 NNUE evaluation.
4. NNUE code to excecute the Benjamin NN evaluation by Chris Whittington.
NNUE evaluation, how it works, steps,
1. Playing millions of fast Benjamin 1.1 games.
2. Creating EPD's from the games.
3. Pruning the EPD's throwing out tactical positions.
4. Making sure you have enough positions left, typically 850 million for Rebel.
5. Put the 850 million into the learner which will create the nnue file.
The result is a blueprint of Benjamin's evaluation and playing style, only better and much stronger.
_____________________________________________________________________________________________________
Playing style, wild and exciting as the title of the song.
At times a bit too aggressive making an unsound move. But Benjamin is the gambit version of the more balanced ProDeo 3.1 after all and so it is tolerable for a first version.
Next versions will focus on the playing style of ProDeo 3.1, there is a whole new world to explore anyway.
Results : 2 matches against an elo pool of 3100 elo rated engines.
Engine | Games | TC | Result | % | Elo | Remark |
Rebel 14 | 1000 | 40/40 | 562.5/1000 | 56.2% | 3136 | Gambit Positions GRL |
Rebel 14 | 1000 | 40/120 | 598.5/1000 | 59.8% | 3159 | Gambit Positions GRL |
This would put Rebel 14 on place 25 of the current Gambit Rating List.
______________________________________________________________________________________________________
Credits
______________________________________________________________________________________________________
• TOGA IV now with Rebel 14.2 NNUE evaluation.
• Toga IV 1.1 is about 45 elo stronger than Rebel 14.2 and above that offers you:
• Multiple threading up to 64 threads;
• MultiPV up to 10 moves.
• TOGA IV 1.1 earned a rating of 3323 on the Gambit Rating List.
• Download TOGA IV 1.1
• Download source code.
• Offered to you in cooperation with Chris Whittington.
______________________________________________________________________________________________________