Proteus Chess SF - developing my Stockfish 14.1 derivative.

Zippo wrote:

AlexChess wrote:

+6 OVER BRAINLEARN 15.2! (That has learning but has never been trained agaist ProteusSF RBE 008b)

DETAILS: https://banksiagui.com/forums/viewtopic.php?p=132#p132

For who could be interested, on BanksiaGui there are also some infos about next ProteusSF on bottom

Just started the challenge to StockfishMZ 170522 as @Zippo asked me here (referring to the version without learning, but this is the same with it, to be fair)

Hi Alex before u go there, u might want to check this out..
mediafire.com file/id28b2gkvn1jeub/Engmatch+9+-1494.pgn/file

It was the PGN's file  from oilprog test, where ur engine lost against EMAN 8.03

Hopely with this PGN file, u can fix whatever u need to fix, either it comes from the EXP FILE ? or the Engine itself ?
I guess u knew it better than me..

Good luck Alex

opening book: OliBiyu ctg is not my suggested book and @Oliprog hasn't used a PGN test suite switching sides for each position like TCEC 22 or DOTS to be fair, as I and @im48 are doing like suggested by openings gurus @dorsz and @sarona , Proteus SelfQ never tested and -2 on a 300 games match against latest Eman 8.03 of my myth Khalid Omar isn't a bad result Next match will be against it  to train ProteusSF against 8.03 and then I'll release an updated ProteusSF.exp

Congratulations Khalid Omar! And thank you for your sublime learning code.

A SNEAK PEEK FOR MY FRIENDS HERE:

HALF WAY


ProteusSF RBE 008b + 5

ALL 50 PGN GAMES UNTIL NOW: https://pixeldrain.com/u/EwFKveu7 (Please analyze them with me)

FOLLOW ME on BanksiaGui Forum Official ProteusSF DEV page!

https://banksiagui.com/forums/viewtopic.php?t=56

In my first tests Proteus 8b did very well. All engines have default settings.



[pgn][Event "proteus8b, Blitz 3.0min+3.0sec"]
[Site "?"]
[Date "2022.06.07"]
[Round "2.8"]
[White "ProteusSF RBE 008b"]
[Black "Fat Fritz 2"]
[Result "1-0"]
[ECO "B30"]
[Annotator "0.25;0.26"]
[PlyCount "101"]
[EventDate "2022.06.07"]
[EventType "simul"]
[SourceTitle "Fritz Engine Tournament"]
[Source "Santos"]
[TimeControl "180+3"]

{11th Gen Intel(R) Core(TM) i7-11390H @ 3.40GHz 2918 MHz W=25.4 plies; 2.292kN/
s B=28.0 plies; 3.197kN/s} 1. e4 {0.25/30 23} c5 {0.26/31 14} 2. Nf3 {0.39/31
13} Nc6 {0.34/28 2} 3. Bb5 {0.43/30 15 (Nc3)} Nf6 {0.35/29 15 (g6)} 4. Bxc6 {
0.66/25 5} dxc6 {0.31/25 2} 5. d3 {0.72/29 3} Qc7 {0.22/25 4 (g6)} 6. e5 {
0.76/27 9 (Nbd2)} Nd5 {0.11/26 4} 7. c4 {0.78/25 2} Nb4 {0.13/29 22} 8. a3 {
0.72/28 5} Na6 {0.21/26 2} 9. O-O {0.57/30 5} Bg4 {0.24/27 5 (Bf5)} 10. Nbd2 {
1.09/26 10 (Re1)} O-O-O {0.28/28 7} 11. Re1 {1.09/29 23 (Ne4)} Rxd3 {0.08/26 3
(Bf5)} 12. h3 {1.00/27 8 (Qe2)} Bxf3 {0.19/29 6 (Bf5)} 13. Qc2 {1.07/28 3} Rd7
{0.06/29 5 (Rd8)} 14. e6 {2.20/25 5 (Nxf3)} Rxd2 {0.00/29 9 (Bh5)} 15. Qxd2 {
3.08/25 5} Bh5 {0.05/27 3} 16. Qg5 {3.44/26 6 (b4)} Bg6 {1.61/32 13} 17. Bf4 {
3.93/27 7} f6 {2.63/32 20} 18. Qg3 {3.87/25 7 (Qg4)} Qb6 {0.69/26 6 (Qa5)} 19.
Qe3 {4.28/25 7 (Rad1)} Bh5 {3.05/34 45} 20. g4 {4.95/25 8} g5 {3.09/32 5} 21.
Bh2 {4.96/24 6 (Bg3)} Bg6 {3.29/29 8} 22. Rad1 {5.02/25 7} Bg7 {3.36/29 4} 23.
Rd7 {5.25/26 10} Re8 {3.25/30 6} 24. Qd2 {5.42/23 6} h6 {3.79/33 33 (Bh8)} 25.
Qc3 {5.96/23 8 (b4)} Nb8 {1.57/20 2 (Bh8)} 26. Red1 {7.28/22 6 (Rc7+)} Nxd7 {
1.35/19 1 (Bh8)} 27. Rxd7 {9.15/26 5 (exd7+)} f5 {7.78/24 4} 28. Qxg7 {9.36/27
7} f4 {8.19/24 3} 29. Qxg6 {9.51/29 8} Rd8 {8.21/23 2} 30. Qxh6 {9.58/27 5}
Rxd7 {8.28/22 1} 31. exd7+ {9.59/25 6} Kxd7 {8.44/21 3} 32. Qxg5 {9.68/26 7}
Qxb2 {8.43/21 3} 33. Qf5+ {9.84/21 7 (Qxf4)} Ke8 {6.98/19 1} 34. Bxf4 {10.39/
22 12} Qf6 {7.90/25 5} 35. Qxf6 {10.60/21 5 (Qe4)} exf6 {9.79/22 4} 36. Be3 {
10.75/20 3 (Kg2)} Kf7 {9.59/27 11 (f5)} 37. Bxc5 {11.00/21 3} b6 {10.01/24 3}
38. Kg2 {11.45/20 3} Ke6 {11.99/23 6} 39. Kf3 {12.09/19 5} f5 {12.97/24 4 (Kd7)
} 40. gxf5+ {12.74/20 3 (g5)} Kxf5 {8.17/17 0} 41. Be3 {13.03/19 3 (Bxb6)} Kg6
{13.15/24 5 (Ke6)} 42. Kg4 {13.86/18 2} Kg7 {14.26/24 2 (Kf6)} 43. Kf5 {
14.50/17 3 (h4)} Kh7 {10.32/25 3 (Kf7)} 44. h4 {20.75/19 6} b5 {99.19/33 5
(Kg8)} 45. cxb5 {43.76/22 2} cxb5 {#19/33 1} 46. Bxa7 {#17/31 3} Kh6 {#15/39 1}
47. Kg4 {#15/33 3} Kg6 {#14/39 1} 48. h5+ {#14/31 4} Kf6 {#13/42 1} 49. Be3 {
#13/31 4 (Bd4+)} Ke5 {#12/41 1} 50. Kg5 {#11/30 2} b4 {#11/45 1 (Ke6)} 51. axb4
{#10/32 3} 1-0

[/pgn]

adaocand wrote:

In my first tests Proteus 8b did very well. All engines have default settings.



[pgn][Event "proteus8b, Blitz 3.0min+3.0sec"]
[Site "?"]
[Date "2022.06.07"]
[Round "2.8"]
[White "ProteusSF RBE 008b"]
[Black "Fat Fritz 2"]
[Result "1-0"]
[ECO "B30"]
[Annotator "0.25;0.26"]
[PlyCount "101"]
[EventDate "2022.06.07"]
[EventType "simul"]
[SourceTitle "Fritz Engine Tournament"]
[Source "Santos"]
[TimeControl "180+3"]

{11th Gen Intel(R) Core(TM) i7-11390H @ 3.40GHz 2918 MHz W=25.4 plies; 2.292kN/
s B=28.0 plies; 3.197kN/s} 1. e4 {0.25/30 23} c5 {0.26/31 14} 2. Nf3 {0.39/31
13} Nc6 {0.34/28 2} 3. Bb5 {0.43/30 15 (Nc3)} Nf6 {0.35/29 15 (g6)} 4. Bxc6 {
0.66/25 5} dxc6 {0.31/25 2} 5. d3 {0.72/29 3} Qc7 {0.22/25 4 (g6)} 6. e5 {
0.76/27 9 (Nbd2)} Nd5 {0.11/26 4} 7. c4 {0.78/25 2} Nb4 {0.13/29 22} 8. a3 {
0.72/28 5} Na6 {0.21/26 2} 9. O-O {0.57/30 5} Bg4 {0.24/27 5 (Bf5)} 10. Nbd2 {
1.09/26 10 (Re1)} O-O-O {0.28/28 7} 11. Re1 {1.09/29 23 (Ne4)} Rxd3 {0.08/26 3
(Bf5)} 12. h3 {1.00/27 8 (Qe2)} Bxf3 {0.19/29 6 (Bf5)} 13. Qc2 {1.07/28 3} Rd7
{0.06/29 5 (Rd8)} 14. e6 {2.20/25 5 (Nxf3)} Rxd2 {0.00/29 9 (Bh5)} 15. Qxd2 {
3.08/25 5} Bh5 {0.05/27 3} 16. Qg5 {3.44/26 6 (b4)} Bg6 {1.61/32 13} 17. Bf4 {
3.93/27 7} f6 {2.63/32 20} 18. Qg3 {3.87/25 7 (Qg4)} Qb6 {0.69/26 6 (Qa5)} 19.
Qe3 {4.28/25 7 (Rad1)} Bh5 {3.05/34 45} 20. g4 {4.95/25 8} g5 {3.09/32 5} 21.
Bh2 {4.96/24 6 (Bg3)} Bg6 {3.29/29 8} 22. Rad1 {5.02/25 7} Bg7 {3.36/29 4} 23.
Rd7 {5.25/26 10} Re8 {3.25/30 6} 24. Qd2 {5.42/23 6} h6 {3.79/33 33 (Bh8)} 25.
Qc3 {5.96/23 8 (b4)} Nb8 {1.57/20 2 (Bh8)} 26. Red1 {7.28/22 6 (Rc7+)} Nxd7 {
1.35/19 1 (Bh8)} 27. Rxd7 {9.15/26 5 (exd7+)} f5 {7.78/24 4} 28. Qxg7 {9.36/27
7} f4 {8.19/24 3} 29. Qxg6 {9.51/29 8} Rd8 {8.21/23 2} 30. Qxh6 {9.58/27 5}
Rxd7 {8.28/22 1} 31. exd7+ {9.59/25 6} Kxd7 {8.44/21 3} 32. Qxg5 {9.68/26 7}
Qxb2 {8.43/21 3} 33. Qf5+ {9.84/21 7 (Qxf4)} Ke8 {6.98/19 1} 34. Bxf4 {10.39/
22 12} Qf6 {7.90/25 5} 35. Qxf6 {10.60/21 5 (Qe4)} exf6 {9.79/22 4} 36. Be3 {
10.75/20 3 (Kg2)} Kf7 {9.59/27 11 (f5)} 37. Bxc5 {11.00/21 3} b6 {10.01/24 3}
38. Kg2 {11.45/20 3} Ke6 {11.99/23 6} 39. Kf3 {12.09/19 5} f5 {12.97/24 4 (Kd7)
} 40. gxf5+ {12.74/20 3 (g5)} Kxf5 {8.17/17 0} 41. Be3 {13.03/19 3 (Bxb6)} Kg6
{13.15/24 5 (Ke6)} 42. Kg4 {13.86/18 2} Kg7 {14.26/24 2 (Kf6)} 43. Kf5 {
14.50/17 3 (h4)} Kh7 {10.32/25 3 (Kf7)} 44. h4 {20.75/19 6} b5 {99.19/33 5
(Kg8)} 45. cxb5 {43.76/22 2} cxb5 {#19/33 1} 46. Bxa7 {#17/31 3} Kh6 {#15/39 1}
47. Kg4 {#15/33 3} Kg6 {#14/39 1} 48. h5+ {#14/31 4} Kf6 {#13/42 1} 49. Be3 {
#13/31 4 (Bd4+)} Ke5 {#12/41 1} 50. Kg5 {#11/30 2} b4 {#11/45 1 (Ke6)} 51. axb4
{#10/32 3} 1-0

[/pgn]

Thank you for the tests, even on mine, is doing incredibly well. But I'm challenging StockfishMZ 170523. Until now -6 66/100 games played on my TCEC 22 match. No match lost until now using this unbalanched opening test suite  



....Tomorrow final result and then Eman 8.03 challenge

Congratulations on your work, we who are chess engine enthusiasts thank you for your dedication. I liked your idea and ran a test of Preteus 8b against Eman 8.3 and MZ170522. All very even, Proteus handled it well. All engines have standard configurations.  

adaocand wrote:

Congratulations on your work, we who are chess engine enthusiasts thank you for your dedication. I liked your idea and ran a test of Preteus 8b against Eman 8.3 and MZ170522. All very even, Proteus handled it well. All engines have standard configurations.  

Thank you! Which openings are you using? Have you added also ProteusSF.exp contained on my zip and Eman.exp from here? https://pixeldrain.com/u/rvhwkwCL I'm about yo start TCEC 22 opening test suite Eman 8.03 vs Proteus RBE 008b

Next TCEC 22 match against Eman 8.03 on Windows 11 ARM64 2Mn/s

Last edited by AlexChess on Thu Jun 09, 2022 3:35 am; edited 2 times in total

@Oliprog @im48 @RSilvian @dorsz @mehmet1921 @adaocand When comparing engines and not books maybe we should select same openings like Perfect or SuperGM_4mvs or better unbalanced PGN test suites switching colors and repeating each position) or strictly use bin books and exp suggested by engine authors. Doing that I and @im48 (Ghppn) we are obtaining very similar results.



ProteusSF RBE 008b +9 https://pixeldrain.com/u/Xu8N1NaL

Last edited by AlexChess on Thu Jun 09, 2022 3:33 am; edited 1 time in total

In the initial tests I prefer to put the engines in standart mode, without any additives or even determining the opening positions. When I use opening books I can't know if the book is the determinant or the algorithm's calculation precision. Yes, I used the latest .exp file in Proteus8b, and I will use this new file .exp from Eman in the next tests. Tks

Instead of wasting electricity on meaningless 20 games, you could also roll dice and the same result would be the outcome.

Fire_Fox wrote:

Instead of wasting electricity on meaningless 20 games, you could also roll dice and the same result would be the outcome.

Waste of electricity and time is you coming here talking shit. Run your tests and present your results. Show factually that the sample size will change the performance of the engines. Only criticizing those who do is always the hobby of the wicked and poor in spirit. And have a good day!

Fire_Fox wrote:

Instead of wasting electricity on meaningless 20 games, you could also roll dice and the same result would be the outcome.

Don't be rude here, we hate haters

adaocand wrote:

In the initial tests I prefer to put the engines in standart mode, without any additives or even determining the opening positions. When I use opening books I can't know if the book is the determinant or the algorithm's calculation precision. Yes, I used the latest .exp file in Proteus8b, and I will use this new file .exp from Eman in the next tests. Tks

When you will test them with books, for fair results please use Solista_010622.bin (contained in the zip) for ProteusSF if against other customized books, otherwise is OK every book you like but playing every position sequentially on both sides (only PGN | EPD books allow that)
With PGN unbalanced test suites is more funny, because there are a lot of win games not 99% draws, but since they play the same positions, the match is fair.



ALL 25/100 GAMES UNTIL NOW: https://pixeldrain.com/u/wNibpFyK +1

[quote="AlexChess"]

Don't be rude here, we hate haters :sm022:

[/qThe truth hurtsuote]

edit:AlexChess wrote "Don't be rude here, we hate haters"
but
The truth hurts

Edit

Last edited by deeds on Fri Jun 10, 2022 6:35 pm; edited 1 time in total

I rolled twenty times today and Proteus won 10:9 over Eman 8.03, but to be fair with the left hand. With the right hand Eman decided the match with 9:10 for his side.

Edit

Last edited by deeds on Fri Jun 10, 2022 6:35 pm; edited 1 time in total

Android for droidfish app Proteus SF testing engine ..just one question.. nnue support enable / desable ?

AlexChess wrote:

adaocand wrote:

In the initial tests I prefer to put the engines in standart mode, without any additives or even determining the opening positions. When I use opening books I can't know if the book is the determinant or the algorithm's calculation precision. Yes, I used the latest .exp file in Proteus8b, and I will use this new file .exp from Eman in the next tests. Tks

When you will test them with books, for fair results please use Solista_010622.bin (contained in the zip) for ProteusSF if against other customized books, otherwise is OK every book you like but playing every position sequentially on both sides (only PGN | EPD books allow that)
With PGN unbalanced test suites is more funny, because there are a lot of win games not 99% draws, but since they play the same positions, the match is fair.

ALL 25/100 GAMES UNTIL NOW: https://pixeldrain.com/u/wNibpFyK +1

"but playing every position sequentially on both sides (only PGN | EPD books allow that)"

How can I use pgn/epd in BanksiaGUI for matches/tournaments?
I can't find that option...

Thank you in advance.

There are four steps:
1. Click Tournament
2. Click New Tour
3. Click on the setup box under Openings:
4. add your pgn

ZamChess wrote:

AlexChess wrote:

adaocand wrote:

In the initial tests I prefer to put the engines in standart mode, without any additives or even determining the opening positions. When I use opening books I can't know if the book is the determinant or the algorithm's calculation precision. Yes, I used the latest .exp file in Proteus8b, and I will use this new file .exp from Eman in the next tests. Tks

When you will test them with books, for fair results please use Solista_010622.bin (contained in the zip) for ProteusSF if against other customized books, otherwise is OK every book you like but playing every position sequentially on both sides (only PGN | EPD books allow that)
With PGN unbalanced test suites is more funny, because there are a lot of win games not 99% draws, but since they play the same positions, the match is fair.

ALL 25/100 GAMES UNTIL NOW: https://pixeldrain.com/u/wNibpFyK +1

"but playing every position sequentially on both sides (only PGN | EPD books allow that)"

How can I use pgn/epd in BanksiaGUI for matches/tournaments?
I can't find that option...

Thank you in advance.

Thank you @Zamchess is nice to have you on our side. On tournament verify to check Play all games sequentially and next 2 for every positions

Alex https://www.chess2u.com/t18259-another-experiment#117001

Krunal wrote:

Android for droidfish app Proteus SF testing engine ..just one question.. nnue support enable / desable ?

ENABLE

But remember to repeat the test a thousand times or Chrome will come here to throw the dice on you!  

[quote="adaocand"]But remember to repeat the test a thousand times or Chrome will come here to throw the dice on you!  [img]https://twemoji.maxcdn.com/16x16/1f631.png[/img][/quote]
I will roll the dice a thousand times when you have played a thousand games and not twenty :sm03: