Trying a more aggressive Rebel-Extreme

[Rebel]

After several failed tries to create an even more aggressive Rebel-Extreme 1.1 the C5x neural nets show improvement.

. v1 is about 800M positions.
. v2 is about 1.6B positions.

In the make is v3 which will take about 10 days and will contain ~2.6B positions.

​New results will be posted in the tournament and matches forum.

For a comparison with the Stephan Pohl EAS rating list check here.

https://rebel7775.wixsite.com/rebel/pla ... ating-list

[pohl4711]

After several failed tries to create an even more aggressive Rebel-Extreme 1.1 the C5x neural nets show improvement.

. v1 is about 800M positions.
. v2 is about 1.6B positions.

In the make is v3 which will take about 10 days and will contain ~2.6B positions.

​New results will be posted in the tournament and matches forum.

For a comparison with the Stephan Pohl EAS rating list check here.

https://rebel7775.wixsite.com/rebel/pla ... ating-list

Looking forward, to test this new one for my full ratinglist and see the EAS-score in my full ratiglist...

Stefan (not Stephan!)

[Rebel]

Fixed it!

[Rebel]

v3 is about 2.6B positions.

Result of v3 is not great except it gains +18 elo over v2.

Next v4, adding 200M balanced positions, sometimes works, sometimes not.

Takes 2-3 days for results.

https://rebel7775.wixsite.com/rebel/pla ... ating-list

[chessica]

Code: Select all

Gemittelt pro Engine
Rank 	Engine 	Elo 	EloError 	EloCI95 	Score 	ScoreRate
1 	Esox-bmi2 	3218 	3.27 	6.41 	13584 	0.906
2 	Esox-avx2 	3216 	3.27 	6.41 	13571 	0.905
3 	Esox-sse41-popcnt 	3210 	3.27 	6.41 	13536 	0.902
4 	stockfish_25120311_x64_avx2 	3196 	3.27 	6.41 	13423 	0.895
5 	stockfish-17.1-avx2 	3194 	3.27 	6.41 	13405 	0.894
6 	DON-BMI2 	3180 	3.27 	6.41 	13284 	0.886
7 	Leptir TI MPV PB-bmi2 	3178 	3.27 	6.41 	13282 	0.885
8 	DON-AVX2 	3178 	3.27 	6.41 	13274 	0.885
9 	DON-BMI2-COMP 	3178 	3.27 	6.41 	13270 	0.885
10 	Leptir TI MPV PB-avx2 	3175 	3.27 	6.41 	13238 	0.883
11 	DON-SSE41 	3173 	3.27 	6.41 	13232 	0.882
12 	Leptir TI MPV PB-sse41 	3168 	3.27 	6.41 	13179 	0.879
13 	reckless-0.9.0 	3164 	3.27 	6.41 	13138 	0.876
14 	reckless_090_dev-162f_64_ja_avx512 	3151 	3.27 	6.41 	13015 	0.868
15 	Brainlearn25.2-x86-64 	3147 	3.27 	6.41 	12973 	0.865
16 	Brainlearn25.2-x86-64-sse3-popcnt 	3136 	3.27 	6.41 	12867 	0.858
17 	PlentyChess-7.0.35 	3131 	3.27 	6.41 	12807 	0.854
18 	Quanticade-Windows-clang-x86-64-bmi2 	3119 	3.27 	6.41 	12679 	0.845
19 	Quanticade-Windows-clang-x86-64-avx512 	3117 	3.27 	6.41 	12654 	0.844
20 	Brainlearn25.2-general-64 	3103 	3.27 	6.41 	12493 	0.833
21 	Brainlearn25.2-x86-32 	3102 	3.27 	6.41 	12487 	0.832
22 	Sun 15 MPV-avx2 	3082 	3.27 	6.41 	12227 	0.815
23 	Sun 15-sse41-popcnt 	3076 	3.27 	6.41 	12151 	0.810
24 	Sun 15-bmi2 	3075 	3.27 	6.41 	12130 	0.809
25 	Brainlearn25.2-general-32 	3069 	3.27 	6.41 	12064 	0.804
26 	Rebel-Extreme-1.1 	3050 	3.27 	6.41 	11793 	0.786
27 	Vajolet2_3.0_bmi 	3008 	3.27 	6.41 	11133 	0.742
28 	senpai_30_64_ja_avx 	2996 	3.27 	6.41 	10930 	0.729
29 	senpai_30_64_ja_avx512 	2995 	3.27 	6.41 	10920 	0.728
30 	senpai_30_64_ja_bmi2 	2995 	3.27 	6.41 	10918 	0.728
31 	Vajolet2_3.0 	2994 	3.27 	6.41 	10906 	0.727
32 	senpai_30_64_ja_sse4 	2994 	3.27 	6.41 	10902 	0.727
33 	senpai_30_64_ja_avx2_zen2 	2993 	3.27 	6.41 	10897 	0.726
34 	senpai_30_64_ja_sse3_nopop 	2992 	3.27 	6.41 	10865 	0.724
35 	Deuterium_v2019.2.37.73_64bit_pop 	2933 	3.27 	6.41 	9778 	0.652

[Rebel]

@chessica, this is a thread about the development of a new PLAYING STYLE engine, it's not about playing strength. Please don't contaminate my thread with off topic info. Thank you.

Available now : Protools-1.8a, it includes how to make the Playing-Style-Rating-List.

https://rebel7775.wixsite.com/rebel/downloads

So @chessica, just try it with the 2 included PGN files for demonstration purposes to get an idea what's this thread is about.

[Rebel]

PGN : pgn\comp-games.pgn

Code: Select all

Playing Style Rating-list  Mon Dec 29 10:56:57 2025

PGN   : pgn\comp-games.pgn
Games : 15000
Won   : 8319
Time  : 5 seconds

 STYLE   King          Short           
 Total  Attack   %     Games   %     Sacs    %    Engine
427707  162561 23.9%  132348 55.2%  132798 35.5%  CSTal 2.1 EAS  
154581   24045 04.0%  125808 62.1%    4728 03.1%  Clover 3.3  
127571   21829 04.3%   98301 60.0%    7441 03.7%  Seer 2.50  
122501   20244 02.3%   95123 61.4%    7134 03.4%  Black Marlin 8.0  
112893   22376 04.0%   84351 56.5%    6166 02.6%  Pawn 3.0  
104153   21506 02.4%   73286 50.0%    9361 04.2%  Wasp 7.0  
 96188   19351 03.4%   70408 43.3%    6429 03.4%  Igel 3.30  
 94264   22986 03.9%   66947 52.0%    4331 02.4%  viridithas 9.0.0  
 93574   15219 02.0%   74003 54.2%    4352 02.4%  Koivisto 6.16  
 93411   17617 02.5%   67784 50.9%    8010 03.7%  Halogen 12  
 83752   18538 02.5%   61217 43.6%    3997 02.3%  Altair 6.0  

[Rebel]

PGN : pgn\human-games.pgn

Code: Select all

Playing Style Rating-list  Mon Dec 29 10:59:53 2025

PGN   : pgn\human-games.pgn
Games : 86129
Won   : 42456
Time  : 7 seconds

 STYLE   King          Short           
 Total  Attack   %     Games   %     Sacs    %    Engine
346113   35336 13.6%  271135 54.3%   39642 15.7%  Svidler, P  
297472   46834 10.4%  205210 38.6%   45428 19.3%  Caruana, F  
291029   34008 07.6%  231293 46.0%   25728 10.9%  So, W  
279238   37120 11.2%  193836 46.2%   48282 17.5%  Aronian, L  
268266   48530 10.7%  197933 46.4%   21803 10.7%  Dominguez Perez, L  
223055   54953 12.4%  108333 40.2%   59769 20.5%  Nepomniachtchi, I  
216985   35768 09.9%  141069 40.1%   40148 15.6%  Vachier Lagrave, M  
198877   46644 12.6%  108561 44.5%   43672 17.2%  Carlsen, M  
182178   33472 08.4%  116699 40.8%   32007 12.4%  Nakamura, Hi  
176761   51559 15.3%   85901 42.1%   39301 12.9%  Firouzja, Alireza  
174263   51292 09.2%   83256 45.9%   39715 15.1%  Dubov, Daniil  
167718   62453 20.0%   45496 50.0%   59769 23.1%  Sutovsky, E.  
166766   45625 11.5%   68097 40.7%   53044 20.4%  Li Chao2  
164091   33908 08.2%   78174 38.1%   52009 16.7%  Fedoseev, Vl3  
161381   55919 11.5%   49490 46.6%   55972 19.8%  Morozevich, A.  
159222   51882 14.2%   54859 35.6%   52481 18.4%  Mamedov, Rau  
157867   37240 10.1%   89651 44.2%   30976 16.3%  Riazantsev, A.  
155807   48685 14.0%   48505 42.7%   58617 21.9%  Andreikin, D.  
154195   50481 12.5%   46232 38.4%   57482 17.9%  Saric, Iv  
153531   54992 12.2%   38862 42.3%   59677 21.0%  Shirov, A.  
152678   54221 11.4%   36271 37.3%   62186 21.7%  Vallejo Pons, F.  
152404   42682 11.5%   61541 46.2%   48181 14.3%  Wei Yi  
149608   64605 18.6%   21978 36.4%   63025 22.0%  Bartel, Mat  
149088   35468 07.8%   82368 43.7%   31252 11.7%  Vidit, S  
146371   51707 13.8%   44021 45.2%   50643 18.8%  Mamedyarov, S.  
144433   45553 10.2%   73169 45.8%   25711 12.4%  Xiong, Jeffery  
144343   60223 13.4%   34208 38.3%   49912 16.8%  Najer, E.  
143860   49519 11.9%   49035 44.6%   45306 18.7%  Ding Liren  
143161   52046 14.3%   41357 45.8%   49758 16.7%  Naiditsch, A.  
142921   49881 10.6%   36983 33.2%   56057 20.9%  Nepomniachtchi, I.  
140690   40523 10.0%   60421 40.9%   39746 15.9%  Carlsen, M.  
140582   47970 11.3%   43584 41.0%   49028 18.0%  Radjabov, T.  
139642   49106 09.7%   42072 46.6%   48464 18.5%  Kramnik, V.  
139231   57686 15.2%   39500 33.3%   42045 18.2%  Cheparinov, I.  
139064   55735 15.9%   23625 40.9%   59704 16.7%  Smirin, I.  
138597   51177 11.6%   40560 49.1%   46860 18.9%  Kasparov, G.  
137567   54553 10.2%   40219 44.6%   42795 18.1%  Mamedyarov, S  
136783   44485 09.4%   46544 35.1%   45754 17.2%  Aronian, L.  
136336   49523 11.3%   38479 41.6%   48334 17.2%  Anand, V.  
135913   44844 09.1%   55198 40.9%   35871 09.8%  Giri, A  
134612   42113 09.6%   46053 38.3%   46446 16.0%  Bacrot, E.  
133941   39853 08.5%   50803 35.1%   43285 18.1%  Dreev, A.  
133195   50439 12.2%   44449 38.3%   38307 13.3%  Caruana, F.  
132465   51124 10.7%   40390 42.7%   40951 15.5%  Movsesian, S.  
132431   37316 08.7%   61751 40.5%   33364 13.9%  So, W.  
131232   42521 10.6%   56296 41.6%   32415 13.3%  Gashimov, V.  
129766   63848 16.2%   36918 39.6%   29000 14.4%  Sokolov, I.  
128763   50738 13.8%   36560 43.1%   41465 16.1%  Rublevsky, S.  
128176   55309 10.3%   23534 31.2%   49333 19.3%  Topalov, V.  
128169   43176 08.9%   46020 41.3%   38973 16.4%  Giri, A.  
128056   45671 12.2%   52864 40.7%   29521 12.5%  Grischuk, A.  
127379   37933 05.1%   45849 38.6%   43597 14.7%  Vachier Lagrave, M.  
126619   51939 16.7%   54125 44.4%   20555 11.1%  Moiseenko, A1.  
126424   28516 06.8%   55022 26.5%   42886 13.6%  Meier, Geo  
125387   35870 11.9%   28417 41.3%   61100 22.9%  Sasikiran, K.  
125105   54284 12.7%   33687 40.0%   37134 14.2%  Short, N.  
124828   63152 17.4%   32031 41.0%   29645 14.9%  Lautier, J.  
124103   47636 07.3%   31217 42.2%   45250 19.5%  Jobava, Ba  
123715   42310 10.6%   41188 31.4%   40217 17.4%  Wojtaszek, R.  
122589   46185 08.9%   41352 35.8%   35052 15.3%  Harikrishna, P.  
120908   59496 07.4%   29305 39.7%   32107 13.2%  Motylev, A.  
120609   48110 08.5%   27325 40.3%   45174 14.7%  Bareev, E.  
120511   43519 11.6%   20604 34.8%   56388 20.8%  Navara, D.  
120429   48829 11.8%   31535 47.3%   40065 19.5%  Korobov, A.  
120112   50133 09.2%   41748 35.8%   28231 10.4%  Kasimdzhanov, R.  
119594   53910 08.8%   44263 51.8%   21421 10.5%  Tkachiev, V.  
118781   52661 12.1%   22954 38.4%   43166 14.6%  Timman, J.  
117235   44504 10.2%   29031 50.4%   43700 19.7%  Korobov, A  
116905   43583 11.2%   37708 40.6%   35614 12.8%  Fressinet, L.  
116484   44198 07.8%   40010 24.8%   32276 12.1%  Alekseev, Evgeny  
115212   42079 12.9%   39093 40.0%   34040 15.6%  Eljanov, P.  
114945   46188 11.3%   46088 35.2%   22669 09.9%  Almasi, Z.  
114189   47457 13.1%   39613 42.6%   27119 10.8%  Duda, J  
113872   47629 09.4%   27321 34.1%   38922 17.5%  Van Wely, L.  
113043   42165 09.4%   38895 41.1%   31983 13.2%  Gelfand, B.  
112561   43012 12.4%   32966 31.8%   36583 15.4%  Karjakin, Sergey  
111444   44241 16.6%   31853 38.9%   35350 14.0%  Volokitin, And  
111076   52364 14.8%   24453 35.2%   34259 14.8%  Polgar, Ju  
110229   38891 09.3%   28067 41.5%   43271 18.6%  Artemiev, V.  
109348   45840 11.3%   20575 26.4%   42933 20.8%  Sargissian, G.  
109303   43260 06.9%   35887 26.0%   30156 13.9%  Wang Yue  
107896   49635 06.7%   26451 35.4%   31810 11.6%  Tomashevsky, E.  
107455   41129 08.9%   33356 31.7%   32970 09.9%  Sevian, Samuel  
106707   39889 09.7%   26243 31.1%   40575 16.8%  Bologan, V.  
106628   38416 11.9%   19699 27.3%   48513 17.5%  Wang Hao  
106578   52644 10.2%   28007 31.4%   25927 11.7%  Kortchnoi, V.  
106532   46007 14.5%   28300 34.3%   32225 12.9%  Yu Yangyi  
106230   42822 11.2%   25374 34.3%   38034 14.0%  Sarana, A  
105647   43689 08.3%   19408 32.2%   42550 12.2%  Bu Xiangzhi  
105412   40410 11.0%   26065 38.7%   38937 13.1%  Artemiev, V  
104616   32998 08.3%   52607 32.0%   19011 08.8%  Malakhov, V.  
104216   43646 11.9%   31739 42.6%   28831 12.5%  Ivanchuk, V.  
103917   40904 11.2%   32068 34.5%   30945 11.2%  Ponomariov, R.  
103505   50323 08.2%   18056 31.0%   35126 17.7%  Beliavsky, A.  
101855   35071 07.8%   24681 34.2%   42103 16.6%  Karpov, A.  
100229   42299 10.2%   22281 43.8%   35649 13.1%  Maghsoodloo, Parham  
 98878   41323 10.1%   29494 41.6%   28061 12.0%  Svidler, P.  
 98306   39092 11.3%   27799 35.8%   31415 14.2%  Jakovenko, D.  
 96296   42800 11.5%   28657 31.1%   24839 11.2%  Adams, Mi  
 95787   31028 10.6%   29538 34.6%   35221 11.5%  Anand, V  
 95065   43225 07.4%   25412 35.4%   26428 12.2%  Dominguez Perez, L.  
 94735   42554 09.7%   31327 37.6%   20854 07.9%  Vitiugov, N.  
 90579   43497 09.4%   20859 33.3%   26223 14.2%  Kamsky, G.  
 90353   30967 10.3%   24353 34.9%   35033 11.6%  Le Quang Liem  
 89017   39336 09.6%   27818 34.8%   21863 11.1%  Inarkiev, E.  
 83881   33420 07.6%   32364 37.0%   18097 06.5%  Grischuk, A  
 82473   41238 12.9%   25528 32.4%   15707 08.2%  Leko, P.  

[chessica]

@chessica, this is a thread about the development of a new PLAYING STYLE engine, it's not about playing strength. Please don't contaminate my thread with off topic info. Thank you.

Available now : Protools-1.8a, it includes how to make the Playing-Style-Rating-List.

https://rebel7775.wixsite.com/rebel/downloads

So @chessica, just try it with the 2 included PGN files for demonstration purposes to get an idea what's this thread is about.

I don't need any lectures from you, thank you. Whether EAS or UHO, to me they're simply distortions of the beautiful game of chess. Johan Cruyff said: "The most difficult path is usually the right one." And not UHO or EAS.

[Rebel]

@chessica, this is a thread about the development of a new PLAYING STYLE engine, it's not about playing strength. Please don't contaminate my thread with off topic info. Thank you.

Available now : Protools-1.8a, it includes how to make the Playing-Style-Rating-List.

https://rebel7775.wixsite.com/rebel/downloads

So @chessica, just try it with the 2 included PGN files for demonstration purposes to get an idea what's this thread is about.

I don't need any lectures from you, thank you. Whether EAS or UHO, to me they're simply distortions of the beautiful game of chess. Johan Cruyff said: "The most difficult path is usually the right one." And not UHO or EAS.

Right.

Continuation of this contaminated thread at : https://www.talkchess.com/viewtopic.php?t=85798

Thread locked.