Release : Chess System Tal EXTREME

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

peter
Posts: 3534
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: Release : Chess System Tal EXTREME

Post by peter »

Rebel wrote: Fri Feb 06, 2026 3:50 pm Can you post one of those 3 epds Fritz crashes, so I can reproduce?
Sorry, no, they all happened in automatic suite- run and I always at once confirmed the error message to make GUI go on wihout losing time for next one position.
Anyhow, here I have a game- pair out of a match against Lc0CPU, (v32.1, net 3395000) single thread, 3'+1", UHO 2024 900-990cp 6mvs (therefore I wanted to play both engines bookless, to make Absurd do so, I had to take out book.bin of the option field in Cutechess GUI, otherwise it used it even with Bookdepth=0, Bookrandom=0).
This one 0-1 of Absurd was in filter of very interesting wins in IWS- Tool, the speciality is, the game with reversed colours was won by Absurd too
:)
[pgn][Event "Abs-Lc0C"]
[Site "?"]
[Date "2026.02.06"]
[Round "203"]
[White "Lc0CPU"]
[Black "Absurd1"]
[Result "0-1"]
[Annotator "peter"]
[ECO "A40"]
[WhiteFideId "-1"]
[BlackFideId "-1"]
[PlyCount "86"]
[GameId "2274567735414838"]
[EventDate "2026.??.??"]
[TimeControl "60/180+1:0"]

1. d4 {book} e6 {book} 2. c4 {book} b6 {book} 3. a3 {book} g6 {book} 4. Nc3 {book} Bg7 {book} 5. e4 {book} Ne7 {book} 6. Nf3 {book} Bb7 {book} 7. h4 {1.05/6 24 3.1s} f5 {-0.93/14 5.1s} (7... h6 {-0.99/6 26} 8. Be3 {0.84/17 3.0s} d6 {-1.01/5 7.2s} 9. Qc2 {0.79/16 3.3s} Nd7 {-1.05/5 4.3s} 10. Be2 {0.88/17 5.0s} Nf6 {-0.88/3 5.1s} 11. h5 {0.91/17 3.8s} Nxh5 {-0.35/4 4.4s} 12. O-O-O {1.21/16 4.1s} Nf6 {-0.34/4 4.2s} 13. d5 {1.14/17 9.8s} e5 {-0.39/5 4.9s} 14. Nh4 {1.11/16 13} Bc8 {0.00/3 4.5s} 15. f3 {1.12/15 9.6s} Nh5 {-0.26/4 4.3s} 16. Bd3 {0.90/15 4.6s} Ng3 {-0.07/4 4.7s} 17. Rh2 {0.98/15 3.4s} Bf6 {0.45/5 4.2s} 18. f4 {1.31/16 2.8s} exf4 {0.02/4 4.6s} 19. Bxf4 {1.52/18 5.5s} Nh5 {0.00/6 4.3s} 20. Be3 {1.31/17 4.6s} Be5 {-0.12/6 4.4s} 21. Nf3 {1.70/17 3.1s} Bxh2 {-0.01/6 4.3s} 22. Nxh2 {1.78/18 3.7s} g5 {0.03/6 4.3s} 23. e5 {1.84/17 3.0s} dxe5 {-0.16/6 4.7s} 24. Nf3 {1.67/16 2.1s} Nf4 {-0.52/6 4.8s} 25. Bxf4 {1.79/18 3.0s} exf4 {-1.42/5 4.4s} 26. Ne4 {2.15/16 3.4s} f5 {-1.31/6 4.4s} 27. Qc3 {4.09/18 10} O-O {0.80/8 4.4s} 28. Rh1 {4.31/18 2.5s} fxe4 {-1.82/4 4.7s} 29. Rxh6 {3.90/18 3.0s} Bf5 {-2.87/5 4.7s} 30. Qh8+ {#21/34 1.3s} Kf7 {-7.69/1 0.003s} 31. Rf6+ {#19/42 2.2s} Ke8 {-104.12/1 0.008s} 32. Rxf8+ {#17/47 2.1s} Kd7 {-108.36/1 0.007s} 33. Ne5+ {#15/54 2.7s} Kc8 {-103.83/5 14} 34. Rxd8+ {#13/58 2.4s} Kb7 {-111.67/1 0.018s} 35. Rxa8 {#11/63 2.3s} Bc8 {-114.56/4 12} 36. Nc6 {#7/65 0.86s} Kxa8 {-120.26/4 4.0s} 37. Nxe7 {#5/116 0.42s} a5 {-#4/2 1.1s} 38. Qxc8+ {#3/116 0.017s} Ka7 {-#2/1 0.003s} 39. Nc6# {#1/116;1-0 Absurd1-Lc0CPU Abs-Lc0C 2026 (203) 0.005s, White mates}) 8. e5 {1.58/4 9.6s} Bxf3 {-0.93/18 8.0s} 9. gxf3 {1.70/5 4.5s} Nbc6 {-0.89/18 7.0s} 10. h5 {2.34/4 4.5s} O-O {-0.83/19 11} 11. hxg6 {2.53/5 4.4s} f4 {-0.74/17 6.6s} 12. gxh7+ {2.82/5 4.1s} Kh8 {-0.68/16 3.9s} 13. Ne2 {3.27/5 4.1s} d6 {-0.79/17 3.2s} 14. Nxf4 {4.47/5 4.7s} dxe5 {-0.25/16 5.3s} 15. Nxe6 {3.38/6 4.3s} Qd6 {-0.75/16 10} 16. Nxf8 {4.18/6 4.6s} Rxf8 {-0.97/15 4.7s} 17. Bd3 {5.41/6 4.3s} Qd7 {-0.86/16 7.8s} 18. d5 {6.26/3 5.0s} Nd4 {-1.20/16 2.7s} 19. Be4 {9.65/5 4.0s} b5 {-1.21/17 4.3s} 20. b3 {8.99/4 4.7s} Nef5 {-0.88/16 7.3s} 21. c5 {6.50/4 4.4s} Ne7 {-0.87/17 5.2s} 22. c6 {4.26/4 4.6s} Qd6 {-1.29/17 3.3s} 23. Rg1 {3.27/4 4.4s} Nxd5 {-1.34/15 8.1s} 24. Qd3 {3.42/4 4.5s} Rd8 {-1.39/18 9.8s} 25. Bg5 {9.28/3 4.5s} Nf4 {-1.72/18 6.8s} 26. Bxf4 {5.09/4 4.3s} exf4 {-1.61/20 11} 27. Rd1 {8.61/4 4.3s} Qh6 {-1.92/15 2.9s} 28. Qxb5 {4.55/3 4.5s} Nxf3+ {0.00/23 2.3s} 29. Bxf3 {0.91/5 4.2s} Bc3+ {0.00/24 2.7s} 30. Rd2 {0.90/6 4.6s} Rxd2 {0.00/20 2.1s} 31. Qc4 {0.02/4 4.4s} Rc2+ {2.66/18 2.5s} 32. Kd1 {-2.71/5 4.5s} Qd6+ {3.25/18 5.0s} 33. Bd5 {-1.45/8 4.1s} Rd2+ {3.46/19 4.1s} 34. Ke1 {-2.25/9 3.9s} Rd4+ {3.70/20 3.4s} 35. Qxc3 {-3.21/7 4.6s} Qe5+ {5.27/19 3.0s} 36. Be4 {-4.59/9 4.3s} Qxe4+ {7.43/17 2.2s} 37. Kf1 {-5.76/8 3.8s} f3 {12.06/19 2.1s} 38. Rg3 {-8.25/8 4.9s} Qb1+ {#17/43 1.1s} 39. Qe1 {-5.74/8 2.8s} Rd1 {#15/57 1.4s} 40. Rg8+ {-83.16/6 4.8s} Kxh7 {#11/35 0.20s} 41. Rg7+ {-110.24/4 4.6s} Kxg7 {#5/116 0.033s} 42. Kg1 {-#4/2 0.002s} Qg6+ {#3/116 0.014s} 43. Kh1 {-#2/1 0.007s} Qg2# {#1/116 0.006s, Black mates} 0-1
[/pgn]
The match was won clearly by Lc0CPU of course (Score of Absurd1 vs Lc0CPU: 99 - 319 - 82), yet here are the EAS- ratings out of this single 500 games- match, showing quite a reversed one picture again, even more amazing with so little number of wins at all, compared to those of the opponent:

Code: Select all

                         early           bad  avg.win 
Rank  EAS-Score  sacs    sacs   shorts  draws  moves  Engine/player 
---------------------------------------------------------------------------
   1    669857  52.53%  23.08%  29.29%  02.44%   54   Absurd1  
   2     98483  00.63%  50.00%  09.75%  60.98%   59   Lc0CPU  
-------------------------------------------------------------------
*** Average length of all won games:             58 moves
*** Movelimit for early sac bonus  :             29 moves
Peter.
peter
Posts: 3534
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: Release : Chess System Tal EXTREME

Post by peter »

In meantime I had another one crash in Fritz 20 again, this time together wit the position, an old test positions ouf of ACT5 (nr.49 there),

[d]4n1k1/8/1b1P3P/1qB3Np/6PK/7Q/p2p1p2/8 w - - 0 1
bm 1.h7+
This one even was (at a single one second try) reproducible as for crashing, after first time in automatic suite, now in infinite analysis, again a crash of the benign kind, that clicking Ok at once makes GUI and engine go on working.
Image
Peter.
peter
Posts: 3534
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: Release : Chess System Tal EXTREME

Post by peter »

Next one, a Pervakov- study, HHdbVII (+4830.55b6e8), again crash in automatic suite as well as in infinite analysis:

[d]4k2r/3Rp1p1/1Kp1P3/Pp1P2b1/4R3/P4P2/1p4q1/1Q5r w k - 0 1

bm 1.Rh4

Image
Peter.
User avatar
Rebel
Posts: 7499
Joined: Thu Aug 18, 2011 12:04 pm
Full name: Ed Schröder

Re: Release : Chess System Tal EXTREME

Post by Rebel »

peter wrote: Fri Feb 06, 2026 11:33 pm Next one, a Pervakov- study, HHdbVII (+4830.55b6e8), again crash in automatic suite as well as in infinite analysis:

[d]4k2r/3Rp1p1/1Kp1P3/Pp1P2b1/4R3/P4P2/1p4q1/1Q5r w k - 0 1
bm 1.Rh4
You are right, could reproduce it, thanks.
90% of coding is debugging, the other 10% is writing bugs.
User avatar
AdminX
Posts: 6372
Joined: Mon Mar 13, 2006 2:34 pm
Location: Acworth, GA

Re: Release : Chess System Tal EXTREME

Post by AdminX »

Rebel wrote: Wed Jan 07, 2026 8:30 pm Image

Chess System Tal EXTREME

Chess System Tal goes EXTREME, it has a brand new neural net (pure chess knowledge) based on playing style in the tradition of the old genius of yesteryear such as Morphy, Steinitz, Lasker, Anderssen.  Chess System Tal EXTREME is the successor of Chess System Tal 2.1 EAS and gained about 60.000 style points which puts Chess System Tal EXTREME on top of both playing-style rating lists.

More...
Wow! Nice to see it's still going strong!
"Good decisions come from experience, and experience comes from bad decisions."
__________________________________________________________________
Ted Summers
User avatar
Leto
Posts: 2141
Joined: Thu May 04, 2006 3:40 am
Location: Dune

Re: Release : Chess System Tal EXTREME

Post by Leto »

I know that EAS balances strength and aggressiveness but what is UHO?
User avatar
Rebel
Posts: 7499
Joined: Thu Aug 18, 2011 12:04 pm
Full name: Ed Schröder

Re: Release : Chess System Tal EXTREME

Post by Rebel »

Leto wrote: Tue Feb 10, 2026 6:35 am I know that EAS balances strength and aggressiveness but what is UHO?
Unusual Human Openings.
90% of coding is debugging, the other 10% is writing bugs.
User avatar
Tibono
Posts: 160
Joined: Sat Aug 01, 2015 6:16 pm
Location: France
Full name: Eric Bonneau

Re: Release : Chess System Tal EXTREME

Post by Tibono »

Rebel wrote: Tue Feb 10, 2026 6:40 am
Leto wrote: Tue Feb 10, 2026 6:35 am I know that EAS balances strength and aggressiveness but what is UHO?
Unusual Human Openings.
or U(nbalanced)HO? Well, same family anyway...
User avatar
Rebel
Posts: 7499
Joined: Thu Aug 18, 2011 12:04 pm
Full name: Ed Schröder

Re: Release : Chess System Tal EXTREME

Post by Rebel »

Tibono wrote: Tue Feb 10, 2026 7:20 am
Rebel wrote: Tue Feb 10, 2026 6:40 am
Leto wrote: Tue Feb 10, 2026 6:35 am I know that EAS balances strength and aggressiveness but what is UHO?
Unusual Human Openings.
or U(nbalanced)HO? Well, same family anyway...
Ha ha, yes.
90% of coding is debugging, the other 10% is writing bugs.
User avatar
pohl4711
Posts: 2910
Joined: Sat Sep 03, 2011 7:25 am
Location: Berlin, Germany
Full name: Stefan Pohl

Re: Release : Chess System Tal EXTREME

Post by pohl4711 »

Leto wrote: Tue Feb 10, 2026 6:35 am I know that EAS balances strength and aggressiveness but what is UHO?
My project to prevent computerchess from draw death. New standard for engine development (used in Fishtest and other CPU-Frameworks for testing and developing engines since years now) and used in the well known engine-tournaments (chesscom and TCEC).

Besides the "funny" "comments" in this thread, serious information about UHO (concept and download) can be found on my website:
https://www.sp-cc.de/uho_2024.htm

Worlds first UHO-Rankinglist (normal Elo-calculation and gamepair-calculation) can be found on my main-site:
https://www.sp-cc.de/


Why using non-UHO openings in these days is just statistical nonsense (draw-ratio above 90%), can be seen here:

Code: Select all

Stockfish 16 vs. Torch 1, singlethread, 3min+1sec, 1000 games. Results from Stockfish 16 point of view.

Balanced openings:
SF_8moves_v3  :     1000 (+64,=916,-20),     Score: 52.2%, Elo: +15, Draws: 91.6%
Feobos-6m-v2.1:     1000 (+78,=911,-11),     Score: 53.4%, Elo: +24, Draws: 91.1%

UHO big file, used in Fishtest:
UHO_4060_v3   :     1000 (+375,=470,-155),   Score: 61.0%, Elo: +78, Draws: 47.0%