ProDeo
Would you like to react to this message? Create an account in a few clicks or log in to continue.
ProDeo

Computer Chess
 
HomeHome  CalendarCalendar  Latest imagesLatest images  FAQFAQ  SearchSearch  MemberlistMemberlist  UsergroupsUsergroups  RegisterRegister  Log in  

 

 Playing SF vs Lc0

Go down 
4 posters
AuthorMessage
Admin
Admin
Admin


Posts : 2528
Join date : 2020-11-17
Location : Netherlands

Playing SF vs Lc0 Empty
PostSubject: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 4:46 pm

Playing SF vs Lc0, on average, when a game is won by Lc0 at which game stage did Lc0 decided the game?

1. Middle game ?
2. End game ?
3. Late end game ?
Back to top Go down
http://rebel13.nl/
Chris Whittington




Posts : 1254
Join date : 2020-11-17
Location : France

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 5:12 pm

Admin wrote:
Playing SF vs Lc0, on average, when a game is won by Lc0 at which game stage did Lc0 decided the game?

1. Middle game ?
2. End game ?
3. Late end game ?

How define decided?
Back to top Go down
Admin
Admin
Admin


Posts : 2528
Join date : 2020-11-17
Location : Netherlands

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 5:18 pm

Winning score and winning position, some user elo is needed Idea
Back to top Go down
http://rebel13.nl/
Dio




Posts : 214
Join date : 2021-08-28

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 5:23 pm

Wins against Stockfish are rare even with Lc0.

In my experience, Lc0 tends to win in the middlegame or in the transition to the endgame.

Admin likes this post

Back to top Go down
Chris Whittington




Posts : 1254
Join date : 2020-11-17
Location : France

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 5:32 pm

Admin wrote:
Winning score and winning position, some user elo is needed Idea

Okay, but a scientific analysis requires some sort of trigger condition run past the set of lc0 wins.

If anyone has a good set of lc0-sf games, I’ll analyse it.

Dio likes this post

Back to top Go down
Admin
Admin
Admin


Posts : 2528
Join date : 2020-11-17
Location : Netherlands

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 5:48 pm

Dio likes this post

Back to top Go down
http://rebel13.nl/
mwyoung

mwyoung


Posts : 880
Join date : 2020-11-25
Location : USA

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 8:57 pm

Admin wrote:
Playing SF vs Lc0, on average, when a game is won by Lc0 at which game stage did Lc0 decided the game?

1. Middle game ?
2. End game ?
3. Late end game ?

Lc0 does not win often playing Stockfish. Here is a game played the other day with Lc0 running on a RTX 3070 played at 3m+2s that was a win for Lc0.

In my top 4 engine match that is still running Stockfish has not lost a game in 286 games at a time control of 7m+5s.

[pgn][Event ""]
[Site "*"]
[Date "2023.02.12"]
[Time "19:38:04"]
[Round "1"]
[Board "19"]
[White "Lc0 v0.29.0 nn 809328"]
[Black "Stockfish DEV 20230209"]
[Result "1-0"]
[Termination "mate"]
[ECO "B01"]
[Opening "Scandinavian (centre counter) defence"]
[TimeControl "180+2"]
[PlyCount "123"]

1.e4 d5 2.exd5 {B01: Scandinavian (centre counter) defence} Qxd5
3.Nc3 Qa5 4.d4 Nf6
5.Nf3 c6 6.Bc4 Bf5 {End of opening}
7.Ne5 {+0.4/20 14267 142567 413/435/152} e6 {+0.6/35 11633 118744262 0/860/140} 8.g4 {+0.4/22 329 121921 413/437/150} Bg6 {+0.6/34 4972 52536779 0/877/123}
9.h4 {+0.4/18 325 117172 411/442/147} Nbd7 {+0.6/33 4437 47384174 0/874/126} 10.Nxd7 {+0.4/18 332 111241 407/448/145} Nxd7 {+0.6/31 5290 55881571 0/889/111}
11.h5 {+0.4/16 313 108052 405/453/142} Be4 {+0.6/30 4558 48813818 0/892/108} 12.O-O {+0.4/16 12770 291609 412/452/136} Bd5 {+0.6/30 4077 43992748 0/878/122}
13.Nxd5 {+0.4/15 3046 311283 410/457/133} cxd5 {+0.6/32 4881 52893925 0/910/90} 14.Bd3 {+0.4/14 322 302420 409/460/131} Bd6 {+0.6/33 10718 108996232 0/890/110}
15.a4 {+0.4/13 35612 576708 401/462/137} g6 {+0.6/32 6524 68016093 0/867/133} 16.h6 {+0.4/16 323 175398 396/467/137} O-O {+0.6/31 10417 107455983 0/872/128}
17.Qf3 {+0.4/14 325 68769 399/463/138} f5 {+0.5/31 11467 122510720 0/927/73} 18.Bg5 {+0.4/17 11306 162302 417/460/123} e5 {+0.5/30 3245 36145316 0/950/50}
19.Bb5 {+0.8/25 5046 105094 591/297/112} e4 {+1.4/34 13583 154041433 0/58/942} 20.Qb3 {+0.8/34 333 86579 592/299/109} Nb6 {+1.2/33 5250 58654296 0/214/786}
21.f3 {+1.0/26 5394 162240 630/281/89} a6 {+1.5/33 13056 149488849 0/40/960} 22.Be2 {+1.0/25 326 145125 624/289/87} e3 {+1.9/37 42142 484408894 0/4/996}
23.gxf5 {+1.0/17 308 128618 623/294/83} Rxf5 {+1.7/31 3736 45231850 0/12/988} 24.f4 {+1.1/27 10723 216488 643/286/71} Re8 {+2.1/31 13396 160543457 0/2/998}
25.c3 {+1.1/23 319 181585 650/282/68} Be7 {+2.1/28 5561 66302229 0/1/999} 26.Bg4 {+1.2/26 6515 232072 660/278/62} Bxg5 {+1.9/31 4605 52577239 0/5/995}
27.Bxf5 {+1.2/26 284 191564 656/284/60} Bxh6 {+2.0/30 6311 71967159 0/2/998} 28.Bd3 {+1.2/24 323 177840 651/292/57} Nd7 {+2.2/34 27286 299585097 0/1/999}
29.Rae1 {+1.6/23 15222 237260 724/223/53} Nf6 {+2.6/27 3324 38024050 0/0/1000} 30.Qxb7 {+1.6/24 311 210973 723/226/51} Qd8 {+2.9/27 5998 70925947 0/0/1000}
31.Rf3 {+1.6/23 333 200256 726/226/48} Ne4 {+2.9/24 2798 32752323 0/0/1000} 32.Bxe4 {+4.7/15 16881 197410 907/70/23} dxe4 {+3.1/21 855 9194634 0/0/1000}
33.Rfxe3 {+4.9/15 4663 236498 909/71/20} Qh4 {+3.2/22 1661 19000589 0/0/1000} 34.Qd7 {+4.8/15 8497 211533 903/79/18} Re7 {+2.9/22 1087 12522856 0/0/1000}
35.Qh3 {+5.0/14 3927 205600 905/79/16} Qxh3 {+3.4/22 4076 49186231 0/0/1000} 36.Rxh3 {+8.8/10 11001 185491 946/47/7} Bxf4 {+3.5/19 918 10954707 0/0/1000}
37.Kf2 {+9.9/10 14003 439970 952/42/6} Kg7 {+3.5/22 3280 37961412 0/0/1000} 38.b4 {+9.1/11 10563 275657 949/44/7} e3+ {+3.8/21 2256 27187513 0/0/1000}
39.Ke2 {+18.3/8 3468 52657 976/21/3} h5 {+4.1/21 2050 22865258 0/0/1000} 40.c4 {+25.7/7 5842 80929 984/14/2} Rd7 {+4.3/21 2009 22980578 0/0/1000}
41.Kd3 {+37.3/7 7231 101972 991/8/1} Be5 {+4.3/21 2002 23049574 0/0/1000} 42.d5 {+55.1/6 2397 33392 995/4/1} Bf4 {+4.9/19 2000 21409395 0/0/1000}
43.b5 {+58.2/6 6512 101813 995/4/1} axb5 {+5.1/17 2000 22588930 0/0/1000} 44.axb5 {+53.2/6 2419 36383 995/4/1} Rd8 {+5.5/18 1999 21429144 0/0/1000}
45.b6 {+58.0/6 6792 102228 995/4/1} Rd6 {+5.3/19 2000 22171173 0/0/1000} 46.Ke4 {+57.8/6 5426 69117 995/4/1} Rf6 {+6.2/18 2000 22703356 0/0/1000}
47.c5 {+76.0/6 3574 53899 997/3/0} Bd6 {+6.7/17 1999 24363324 0/0/1000} 48.cxd6 {+101.6/6 3562 46295 999/1/0} Rxd6 {+8.8/15 1999 28828211 0/0/1000}
49.Rb1 {+61.2/6 5210 87976 996/3/1} e2 {+9.5/16 1994 31019514 0/0/1000} 50.Rhh1 {+59.6/6 4335 79258 996/3/1} Rd8 {+9.4/17 2004 31404708 0/0/1000}
51.b7 {+84.3/5 3257 64456 998/2/0} h4 {+10.1/14 2001 33474874 0/0/1000} 52.b8=Q {+54.5/6 4204 68669 996/2/2} Rxb8 {M-10/27 803 8353043 0/0/1000}
53.Rxb8 {+73.5/7 1659 36438 998/1/1} Kf6 {M-8/45 1440 17083984 0/0/1000} 54.Rb6+ {+63.0/6 5576 139767 996/3/1} Kg7 {M-7/51 503 6207152 0/0/1000}
55.Kf4 {+62.3/6 3054 81240 996/3/1} Kf7 {M-6/75 627 8573617 0/0/1000} 56.Re6 {M+111/6 2771 73334 1000/0/0} h3 {M-5/211 854 15306685 0/0/1000}
57.Rxh3 {M+108/1 10 467 1000/0/0} e1=Q {M-5/245 776 24889677 0/0/1000} 58.Rxe1 {M+107/1 14 9 1000/0/0} g5+ {M-4/245 41 2400100 0/0/1000}
59.Kxg5 {M+4/4 160 1305 1000/0/0} Kg7 {M-3/245 24 146339 0/0/1000} 60.Re8 {M+3/1 12 318 1000/0/0} Kf7 {M-2/245 24 76490 0/0/1000}
61.Rhe3 {M+2/1 10 10 1000/0/0} Kg7 {M-1/245 26 3936 0/0/1000} 62.R3e7# {M+1/1 10 2 1000/0/0} 1-0
[/pgn]
Back to top Go down
Chris Whittington




Posts : 1254
Join date : 2020-11-17
Location : France

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 10:14 pm

Admin wrote:
http://ccrl.chessdom.com/ccrl/4040/games.html

CTRL-F + lc0


Well, I found all the Stockfish15 games against Lc0 at CCRL 40/15.

84 games in total
Lc0 Wins 0 Draws 39, Losses 45

So I can't really answer the question, since Lc0 doesn't have any wins.
Any more data, anywhere else?


Back to top Go down
mwyoung

mwyoung


Posts : 880
Join date : 2020-11-25
Location : USA

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 10:28 pm

Chris Whittington wrote:
Admin wrote:
http://ccrl.chessdom.com/ccrl/4040/games.html

CTRL-F + lc0


Well, I found all the Stockfish15 games against Lc0 at CCRL 40/15.

84 games in total
Lc0 Wins 0 Draws 39, Losses 45

So I can't really answer the question, since Lc0 doesn't have any wins.
Any more data, anywhere else?



I doubt it. As Lc0 was no match for Stockfish 15 DEV. Until the development of the new Lc0 t80 nets. The game above is the only win for Lc0 using the new t80 nets.

I have about 100 games played SF vs Lc0 using the new t80 nets at 7m+5s. 98 of the games were draws, and Stockfish winning 2 games.

That is all the data I have with a win for Lc0.
Back to top Go down
Chris Whittington




Posts : 1254
Join date : 2020-11-17
Location : France

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 10:43 pm

mwyoung wrote:
Chris Whittington wrote:
Admin wrote:
http://ccrl.chessdom.com/ccrl/4040/games.html

CTRL-F + lc0


Well, I found all the Stockfish15 games against Lc0 at CCRL 40/15.

84 games in total
Lc0 Wins 0 Draws 39, Losses 45

So I can't really answer the question, since Lc0 doesn't have any wins.
Any more data, anywhere else?



I doubt it. As Lc0 was no match for Stockfish 15 DEV. Until the development of the new Lc0 t80 nets. The game above is the only win for Lc0 using the new t80 nets.

I have about 100 games played SF vs Lc0 using the new t80 nets at 7m+5s. 98 of the games were draws, and Stockfish winning 2 games.

That is all the data I have with a win for Lc0.

So, no answer possible. Ed, how about LC0 against weaker opposition?
Back to top Go down
mwyoung

mwyoung


Posts : 880
Join date : 2020-11-25
Location : USA

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyWed Feb 15, 2023 10:48 pm

Chris Whittington wrote:
mwyoung wrote:
Chris Whittington wrote:
Admin wrote:
http://ccrl.chessdom.com/ccrl/4040/games.html

CTRL-F + lc0


Well, I found all the Stockfish15 games against Lc0 at CCRL 40/15.

84 games in total
Lc0 Wins 0 Draws 39, Losses 45

So I can't really answer the question, since Lc0 doesn't have any wins.
Any more data, anywhere else?





I doubt it. As Lc0 was no match for Stockfish 15 DEV. Until the development of the new Lc0 t80 nets. The game above is the only win for Lc0 using the new t80 nets.

I have about 100 games played SF vs Lc0 using the new t80 nets at 7m+5s. 98 of the games were draws, and Stockfish winning 2 games.

That is all the data I have with a win for Lc0.

So, no answer possible. Ed, how about LC0 against weaker opposition?

Lc0 t80 has 7 wins total so far playing Dragon 3.2 and Berserk 10 from my current match. If Ed wants those games.
Back to top Go down
Admin
Admin
Admin


Posts : 2528
Join date : 2020-11-17
Location : Netherlands

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyThu Feb 16, 2023 8:03 am

It seems lc0 website is down, can't download T80.
Back to top Go down
http://rebel13.nl/
Chris Whittington




Posts : 1254
Join date : 2020-11-17
Location : France

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyThu Feb 16, 2023 8:23 am

mwyoung wrote:
Chris Whittington wrote:
mwyoung wrote:
Chris Whittington wrote:
Admin wrote:
http://ccrl.chessdom.com/ccrl/4040/games.html

CTRL-F + lc0


Well, I found all the Stockfish15 games against Lc0 at CCRL 40/15.

84 games in total
Lc0 Wins 0 Draws 39, Losses 45

So I can't really answer the question, since Lc0 doesn't have any wins.
Any more data, anywhere else?





I doubt it. As Lc0 was no match for Stockfish 15 DEV. Until the development of the new Lc0 t80 nets. The game above is the only win for Lc0 using the new t80 nets.

I have about 100 games played SF vs Lc0 using the new t80 nets at 7m+5s. 98 of the games were draws, and Stockfish winning 2 games.

That is all the data I have with a win for Lc0.

So, no answer possible. Ed, how about LC0 against weaker opposition?

Lc0 t80 has 7 wins total so far playing Dragon 3.2 and Berserk 10 from my current match. If Ed wants those games.

I'll need ALL games (WDL) to determine "Winning eval" level.
Anybody with any more?
Back to top Go down
Admin
Admin
Admin


Posts : 2528
Join date : 2020-11-17
Location : Netherlands

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyThu Feb 16, 2023 8:40 am

https://www.sp-cc.de/nn-vs-kd-testing.htm

Stefan is frequently testing lc0, maybe he can provide a pgn download link.
Back to top Go down
http://rebel13.nl/
Dio




Posts : 214
Join date : 2021-08-28

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyThu Feb 16, 2023 9:44 am

Back to top Go down
mwyoung

mwyoung


Posts : 880
Join date : 2020-11-25
Location : USA

Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 EmptyThu Feb 16, 2023 11:00 am

Chris Whittington wrote:
mwyoung wrote:
Chris Whittington wrote:
mwyoung wrote:
Chris Whittington wrote:
Admin wrote:
http://ccrl.chessdom.com/ccrl/4040/games.html

CTRL-F + lc0


Well, I found all the Stockfish15 games against Lc0 at CCRL 40/15.

84 games in total
Lc0 Wins 0 Draws 39, Losses 45

So I can't really answer the question, since Lc0 doesn't have any wins.
Any more data, anywhere else?





I doubt it. As Lc0 was no match for Stockfish 15 DEV. Until the development of the new Lc0 t80 nets. The game above is the only win for Lc0 using the new t80 nets.

I have about 100 games played SF vs Lc0 using the new t80 nets at 7m+5s. 98 of the games were draws, and Stockfish winning 2 games.

That is all the data I have with a win for Lc0.

So, no answer possible. Ed, how about LC0 against weaker opposition?

Lc0 t80 has 7 wins total so far playing Dragon 3.2 and Berserk 10 from my current match. If Ed wants those games.

I'll need ALL games (WDL) to determine "Winning eval" level.
Anybody with any more?

Here is the tournament games with all Lc0 t80 games.

https://drive.google.com/file/d/1N3fufal9lvvFQh4lt71OnunjjakrsG7z/view?usp=share_link
Back to top Go down
Sponsored content





Playing SF vs Lc0 Empty
PostSubject: Re: Playing SF vs Lc0   Playing SF vs Lc0 Empty

Back to top Go down
 
Playing SF vs Lc0
Back to top 
Page 1 of 1
 Similar topics
-
» Which engines are playing the opening well?
» Benjamin 1.1 - Improved Playing Style and Strength.
» Stockfish demonstrates insane level of playing strength.
» Stockfish 13 Playing Strength Comparison with Grandmasters at 100 milliseconds a move.

Permissions in this forum:You cannot reply to topics in this forum
ProDeo :: Computer Chess-
Jump to: