Hermann 2.0 looks like a nice improvement

[Tony Thomas]

I tested the new version of Hermann just because I didnt test the last two versions. It seems more than a 60 elo improvement compared to the Hermann 1.7 that I tested last year. I thought I should let all of you know that this engine is worth a look.

Code: Select all

3 Hermann 2.0               : 2541  104 (+ 65,=  9,- 30), 66.8 %

AICE 0.99.2                   :   4 (+  1,=  1,-  2), 37.5 %
Anechka 0.08                  :   4 (+  2,=  0,-  2), 50.0 %
Ant 2006F                     :   4 (+  4,=  0,-  0), 100.0 %
Ayito 0.2.994                 :   4 (+  3,=  0,-  1), 75.0 %
Arion 1.7                     :   4 (+  2,=  0,-  2), 50.0 %
Bruja_19                      :   4 (+  3,=  0,-  1), 75.0 %
Delphil 1.6c                  :   4 (+  2,=  0,-  2), 50.0 %
Diablo_04_dc.ex               :   4 (+  4,=  0,-  0), 100.0 %
Djinn 0.925x                  :   4 (+  4,=  0,-  0), 100.0 %
Eeyore 1.48                   :   4 (+  3,=  1,-  0), 87.5 %
EXchess 5.01                  :   4 (+  3,=  0,-  1), 75.0 %
Gaia 3.5                      :   4 (+  1,=  0,-  3), 25.0 %
Horizon 4.3                   :   4 (+  1,=  1,-  2), 37.5 %
Knightx192                    :   4 (+  3,=  1,-  0), 87.5 %
Latista 1.5                   :   4 (+  2,=  1,-  1), 62.5 %
Muse0899b                     :   4 (+  1,=  1,-  2), 37.5 %
Resp_0_1_9                    :   4 (+  1,=  0,-  3), 25.0 %
Scidlet 3.6                   :   4 (+  4,=  0,-  0), 100.0 %
ThorsHammer 2.28              :   4 (+  3,=  0,-  1), 75.0 %
Tornado 1.0 mainz             :   4 (+  4,=  0,-  0), 100.0 %
Twisted Logic 0.65e35         :   4 (+  3,=  1,-  0), 87.5 %
Typhoon 1.00 rev 281          :   4 (+  2,=  0,-  2), 50.0 %
Tytan 8.84                    :   4 (+  2,=  0,-  2), 50.0 %
Xpdnt_060829                  :   4 (+  4,=  0,-  0), 100.0 %
Alfil 6.9                     :   4 (+  1,=  1,-  2), 37.5 %
Danasah 2.85                  :   4 (+  2,=  1,-  1), 62.5 %

Rating of previous version

Code: Select all

14 Hermann 1.7                    : 2465   48  47   173    59.2 %   2400   19.1 %

Just to get a general idea of how strong Hermann is here is a small part of my division 3 rating list from last edition, Hermann is currently # 3.

Code: Select all

5/29/2007 8:50:00 AM :

    Program                          Elo    +   -   Games   Score   Av.Op.  Draws

 1 Alaric v 061008.exe            : 2610   71  69   100    76.0 %   2409   18.0 %
  2 Typhoon 1.00 rev 281           : 2542   50  49   177    69.2 %   2402   17.5 %
  3 Hermann 2.0                    : 2541   69  67   104    66.8 %   2419    8.7 %
  4 Twisted Logic 0.65e35          : 2527   50  49   177    67.2 %   2402   15.8 %
  5 Danasah 2.85                   : 2514   53  53   132    64.4 %   2411   25.8 %
  6 RomiSuperBeyotch1              : 2497   58  57   124    62.5 %   2408   16.9 %
  7 Bruja_19                       : 2495   48  48   180    63.3 %   2400   16.7 %
  8 Gaia 3.5                       : 2491   49  49   177    62.4 %   2403   14.1 %
  9 RomichessP3jBeta1              : 2481   53  53   145    61.7 %   2398   17.2 %
 10 Alfil 6.9                      : 2478   50  49   169    60.7 %   2403   14.8 %
 11 Ayito 0.2.994                  : 2471   49  49   179    60.1 %   2401   11.7 %
 12 Arion 1.7                      : 2469   45  44   180    59.7 %   2401   26.1 %

[Werner]

Hi Tony,
this week I am also testing Hermann 2.0. In the first match against Bruja 1.9 it was clear better. But I stopped this match because I had problems with the output of Bruja. Very often the engine played another move which was not to be seen in the main line!
So I am not sure if Bruja works correct here inside Shredder GUI. I used wb2uci adapter. Now it plays against Gaia 3.5 64 bit. Against Postmodernist it has reached 54%.

[Spock]

I tested the new version of Hermann just because I didnt test the last two versions. It seems more than a 60 elo improvement compared to the Hermann 1.7 that I tested last year. I thought I should let all of you know that this engine is worth a look.

OK - but in my FRC list, Hermann 1.9 is 59 ELO better than Hermann 1.7, so then how much of an improvement I wonder is 2.0 over 1.9....

http://www.computerchess.org.uk/ccrl/40 ... t_all.html

I will test Hermann 2.0 at FRC in the near future

[Tony Thomas]

I tested the new version of Hermann just because I didnt test the last two versions. It seems more than a 60 elo improvement compared to the Hermann 1.7 that I tested last year. I thought I should let all of you know that this engine is worth a look.

OK - but in my FRC list, Hermann 1.9 is 59 ELO better than Hermann 1.7, so then how much of an improvement I wonder is 2.0 over 1.9....

http://www.computerchess.org.uk/ccrl/40 ... t_all.html

I will test Hermann 2.0 at FRC in the near future

I dont think I can help you as I didnt play with Hermann 1.9, may be 20 points is a good bet.

[Tony Thomas]

Hi Tony,
this week I am also testing Hermann 2.0. In the first match against Bruja 1.9 it was clear better. But I stopped this match because I had problems with the output of Bruja. Very often the engine played another move which was not to be seen in the main line!
So I am not sure if Bruja works correct here inside Shredder GUI. I used wb2uci adapter. Now it plays against Gaia 3.5 64 bit. Against Postmodernist it has reached 54%.

Please test Bruja as a winboard engine under Arena, I have yet to see a problem from that fine engine.

[Graham Banks]

Please test Bruja as a winboard engine under Arena, I have yet to see a problem from that fine engine.

Be aware that Bruja has an internal opening book that can't be disabled.

[Tony Thomas]

Please test Bruja as a winboard engine under Arena, I have yet to see a problem from that fine engine.

Be aware that Bruja has an internal opening book that can't be disabled.

Wouldnt the GUI book override the internal book?

[Graham Banks]

Please test Bruja as a winboard engine under Arena, I have yet to see a problem from that fine engine.

Be aware that Bruja has an internal opening book that can't be disabled.

Wouldnt the GUI book override the internal book?

The internal book could still have an influence. This is the reason that CCRL don't test Bruja.

[Tony Thomas]

Please test Bruja as a winboard engine under Arena, I have yet to see a problem from that fine engine.

Be aware that Bruja has an internal opening book that can't be disabled.

Wouldnt the GUI book override the internal book?

The internal book could still have an influence. This is the reason that CCRL don't test Bruja.

I dont think the internal book would have any influence if you use a book that makes the first 10 moves, I doubt that the internal book is any bigger than that.

[Werner]

Thanks, I will have a look at that because I use a book from Salvo. Next i will use my opening positions