OpenChess

Independent Computer Chess Discussion Forum
https://open-chess.org/

deeds_chess_system

https://open-chess.org/viewtopic.php?t=3527

Page 13 of 43

Re: opening book vs experience file

Posted: Sun Jun 25, 2023 7:44 pm
by deeds
I will only open sales twice a year, now each learning session takes several months (= tc 2m+2s).

The new release will be announced on Bettina&Terry77 too.

Re: opening book vs experience file

Posted: Mon Jun 26, 2023 10:09 am
by botunnet
ok i see... No account on morethanchessagame at the moment :oops:

In the era of learning engines, are there still opening books published without a defensive line after g4 d5 :?: :o

:arrow: rnbqkbnr/ppp1pppp/8/3p4/6P1/8/PPPPPP1P/RNBQKBNR w KQkq - 0 2
Image

Image

Re: opening book vs experience file

Posted: Mon Jun 26, 2023 10:24 am
by deeds
botunnet wrote: Mon Jun 26, 2023 10:09 am In the era of learning engines, are there still opening books published without a defensive line after g4 d5 :?: :o
I confess that I don't know, here we haven't checked many books yet, for the moment all the other books have at least one line of attack/defense :

Image

For leschats it's different because its author trains Eman so he knows the learning opportunities.

I regularly update my website so you will see the stats of other books during the year...

Re: opening book vs experience file

Posted: Mon Jun 26, 2023 10:37 am
by deeds
i forgot these ones :

agarta.bin

Code: Select all

a2a3 : .....3 moves
a2a4 : ....31 moves
b2b3 : ....17 moves
b2b4 : .....1 moves
c2c3 : ....74 moves
c2c4 : 10 000 moves
d2d3 : ....75 moves
d2d4 : 10 000 moves
e2e3 : ....84 moves
e2e4 : 10 000 moves
f2f3 : .....0 moves
f2f4 : .....3 moves
g2g3 : ....77 moves
g2g4 : .....2 moves
h2h3 : ...140 moves
h2h4 : .....0 moves
b1a3 : .....1 moves
b1c3 : ....16 moves
g1f3 : 10 000 moves
g1h3 : .....2 moves

Opening Rate : 20%

Hubble 3.1.bin (old version)

Code: Select all

a2a3 : 10 000 moves
a2a4 : .3 732 moves
b2b3 : 10 000 moves
b2b4 : ....84 moves
c2c3 : 10 000 moves
c2c4 : 10 000 moves
d2d3 : ...545 moves
d2d4 : 10 000 moves
e2e3 : 10 000 moves
e2e4 : 10 000 moves
f2f3 : .....0 moves
f2f4 : 10 000 moves
g2g3 : 10 000 moves
g2g4 : .....1 moves
h2h3 : ...271 moves
h2h4 : .....0 moves
b1a3 : ....19 moves
b1c3 : 10 000 moves
g1f3 : 10 000 moves
g1h3 : .....0 moves

Opening Rate : 60%
But hey it's extremely rare nowadays.
I suspect book's authors know they will play against learning engines on online servers so they expand their books.

Re: pawn-value based exp file vs polyglot bin book

Posted: Tue Jun 27, 2023 6:16 am
by deeds
pawn-value based experience file : experience data purely trained by Eman engine until 8.40 release (before Nov 5, 2022 patch)

hardware : dual xeon e5-2660v3
software : cutechess-gui 1.2.0
settings : TC 60s+1s, 40 threads, hash 4096 MB, partial syzygy 7men

Opening list : startpos

Code: Select all

   # PLAYER                           :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L   D(%)  OppAvg  OppN
   1 Depth4_190623.exp (Eman 8.40)    :       0     49    50.0     100  50.0    0  100    0  100.0       0     1
   2 hcanfree-upd.bin (Eman 8.40)     :       0   ----    50.0     100  50.0    0  100    0  100.0       0     1

White advantage = 0.00 +/- 25.05
Draw rate (equal opponents) = 50.00 % +/- 0.00

Opening list : GHPPN's 15 most popular openings (hcanfree-upd's Opening Rate : 27%)

Code: Select all

   # PLAYER                           :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L   D(%)  OppAvg  OppN
   1 Depth4_190623.exp (Eman 8.40)    :       0     92    15.0      30  50.0    0   30    0  100.0       0     1
   2 hcanfree-upd.bin (Eman 8.40)     :       0   ----    15.0      30  50.0    0   30    0  100.0       0     1

White advantage = 0.00 +/- 46.89
Draw rate (equal opponents) = 50.00 % +/- 0.00

Opening list : 20 first legal moves (hcanfree-upd's Opening Rate : 15%)

Code: Select all

   # PLAYER                           :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 Depth4_190623.exp (Eman 8.40)    :      18     81    21.0      40  52.5    2   38    0  95.0       0     1
   2 hcanfree-upd.bin (Eman 8.40)     :       0   ----    19.0      40  47.5    0   38    2  95.0      18     1

White advantage = -17.58 +/- 41.39
Draw rate (equal opponents) = 50.00 % +/- 0.00

Opening list : Albert Silver's 25 gambit suit (hcanfree-upd's Opening Rate : 4%)

Code: Select all

   # PLAYER                           :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 Depth4_190623.exp (Eman 8.40)    :      35     69    27.5      50  55.0    9   37    4  74.0       0     1
   2 hcanfree-upd.bin (Eman 8.40)     :       0   ----    22.5      50  45.0    4   37    9  74.0      35     1

White advantage = 7.08 +/- 35.50
Draw rate (equal opponents) = 50.00 % +/- 0.00

Opening list : Jeroen Noomen's 50 tcec season 22 superfinal openings (hcanfree-upd's Opening Rate : 0%)

Code: Select all

   # PLAYER                           :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 Depth4_190623.exp (Eman 8.40)    :     113     56    64.0     100  64.0   31   66    3  66.0       0     1
   2 hcanfree-upd.bin (Eman 8.40)     :       0   ----    36.0     100  36.0    3   66   31  66.0     113     1

White advantage = 119.59 +/- 29.37
Draw rate (equal opponents) = 50.00 % +/- 0.00
link : games

Re: pawn-value based exp file vs polyglot bin book

Posted: Tue Jun 27, 2023 6:18 am
by deeds
Overall :

Code: Select all

   # PLAYER                           :  RATING  ERROR  POINTS  PLAYED   (%)    W    D    L  D(%)  OppAvg  OppN
   1 Depth4_190623.exp (Eman 8.40)    :      39     27   177.5     320  55.5   42  271    7  84.7       0     1
   2 hcanfree-upd.bin (Eman 8.40)     :       0   ----   142.5     320  44.5    7  271   42  84.7      39     1

White advantage = 32.23 +/- 13.42
Draw rate (equal opponents) = 50.00 % +/- 0.00

Code: Select all

002 players               : avg. D31, 002 125 ms/move, 000 320 games, 00 032 063 moves

depth4_190623 (eman 8.40) : avg. D31, 001 991 ms/move, 000 320 games, 00 017 085 moves
hcanfree-upd (eman 8.40)  : avg. D32, 002 278 ms/move, 000 320 games, 00 014 978 moves

Averages :
0 213 sec/game (4 min/game)
115 plies/game (100 played plies/game)

Game stages :
- opening @ avg. D26, 003 338 ms/move
- middleg @ avg. D29, 001 821 ms/move
- endgame @ avg. D38, 001 253 ms/move

Re: opening book vs experience file

Posted: Tue Jun 27, 2023 8:31 pm
by botunnet
deeds wrote: Mon Jun 26, 2023 10:37 am But hey it's extremely rare nowadays.
I suspect book's authors know they will play against learning engines on online servers so they expand their books.
Completely unusable at school or in a club, kids can't even have fun with wacky openings.

Re: opening book vs experience file

Posted: Tue Jun 27, 2023 10:36 pm
by Sedat Canbaz
------------
Notice:

This post in this thread has been reviewed and edited because it violated the OpenChess Forum Rules. This is a gentle reminder not to continue flaming this thread or others in the Forum. Further recurrent violations of the OpenChess Forum Rules will not be tolerated.

Thanks,

Best regards, Moderation.
------------

deeds wrote: Thu Jun 22, 2023 4:44 pm
Sedat Canbaz wrote: Wed Jun 21, 2023 7:52 pm
Even if they are available...in PGN, CBH etc.
it does not mean that:
Your current played match conditions are fair....

As long as you set learning engines in training mode (default settings) to make your unfair 1vs12 tourneys, you will never have any credibility in learning engines sphere.
By the way,
What about the 3rd Duel Experience vs Book ?
https://sites.google.com/site/computers ... k-vs-exp-3

May I ask you a few but very simple questions,
Then how RapTora's Experience is managed to win ?
Why did you not post screenshots, blame about 3rd duel ?

Re: opening book vs experience file

Posted: Tue Jun 27, 2023 11:07 pm
by Anton101
The only way to get out of doubt if exp file's training that Deeds claims is the way we should follow is that the Deed's experience file enters a tournament vs books (as RapTora did), Using that dev of EMAN engine that Deeds uses for training and, to keep this file without corruption of any kind, the exp file must be NOT updated during the games.

I don't see any other way to fix this. Sedat is undoubtedly professional and we all know he will ensure that the experience file does not fall into other hands.

But the most important thing... Will the training that the file has been subjected will be solid in real games?

It's up to Deeds of course.

Re: opening book vs experience file

Posted: Tue Jun 27, 2023 11:29 pm
by Sedat Canbaz
------------
Notice:

This post in this topic has been reviewed and edited because it violated the OpenChess Forum Rules. This is a gentle reminder not to continue flaming this topic or others in the Forum. Further recurrent violations of the OpenChess Forum Rules will not be tolerated.

Thanks,

Best regards, Moderation.
------------
Anton101 wrote: Tue Jun 27, 2023 11:07 pm The only way to get out of doubt if exp file's training that Deeds claims is the way we should follow is that the Deed's experience file enters a tournament vs books (as RapTora did), Using that dev of EMAN engine that Deeds uses for training and, to keep this file without corruption of any kind, the exp file must be NOT updated during the games.

I don't see any other way to fix this. Sedat is undoubtedly professional and we all know he will ensure that the experience file does not fall into other hands.

But the most important thing... Will the training that the file has been subjected will be solid in real games?

It's up to Deeds of course.
Dear Anton,

Thanks ..

Greetings
All times are UTC+01:00
Page 13 of 43

Powered by phpBB® Forum Software © phpBB Limited