Author: pavel
Date: 02:41:15 05/01/00
Go up one level in this thread
On May 01, 2000 at 05:09:13, blass uri wrote:
>On May 01, 2000 at 03:47:52, pavel wrote:
>
>>On April 30, 2000 at 16:56:30, Christophe Theron wrote:
>>
>>>On April 30, 2000 at 09:51:26, Chessfun wrote:
>>>
>>>>All games on Two Cel 433's.
>>>>Ponder=on.
>>>>Tablebases used at 25 and longer.
>>>>Nunn 1 positions.
>>>>No opening books are loaded.
>>>>
>>>> 1 min game Fritz 6a 18.0 - 2.0 Crafty 17-10
>>>> 2 min game Fritz 6a 17.5 - 2.5 Crafty 17-10
>>>> 3 min game Fritz 6a 15.5 - 4.5 Crafty 17-10
>>>> 5 min game Fritz 6a 15.0 - 5.0 Crafty 17-10
>>>>
>>>>I happened to let 5/0 run through the night.
>>>>As a result it looped through and played 5/0 twice.
>>>>
>>>>In the second set of 20 games the score was:
>>>> 5 min game Fritz 6a 9.0 - 11.0 Crafty 17-10
>>>>I am not sure what to make of it.
>>>
>>>
>>>You should replay the match and first make sure that any learning file is
>>>cleared before restarting.
>>>
>>>However this kind of "strange" result is not unusual and reflects the
>>>variability of chess matches and explains why it is needed to play as many games
>>>as possible.
>>>
>>>
>>> Christophe
>>
>>Hi Christophe
>> i dont get it why would you think the learning file should be
>>cleared for crafty? If so then the same should be done for Fritz as it has
>>as_good_learning_feature_as_crafty. your above statement makes no sense to me
>>;9. And even if crafty wins because of its learning feature, (though i have
>>mentuned lotsa times before that crafty doesnt have it regular position.lrn and
>>book.lrn like the "normal" non_fritz_crafty in under chessbase GUI!!)
>
>
>How do you know that it does not have the regular position.lrn?
>
>I looked at some ssdf games of crafty and while crafty repeated the same winning
>opening against Rebel8(p200) a lot of times because of book learning it did not
>repeat the same games(maybe because of learning by position).
>
>, then it
>>is because it has a better learning code then fritz, which is a part of its
>>palying strength. Why should it be handicapped from it?? Crafty is having enough
>>trouble runnign under fritz GUI, now if it is stoped from learning, then it is
>>as good as making crafty the_way_you_like_it_to_play!! Negative thinking form
>>you and unexpected too :(
>>I really hope reply to my post and make it more clear what you mean :)
>>pavel
>
>The nunn match is not supposed to find the strength of the programs but the
>strength of the engines.
>
>learning by position is not important for regular ssdf games because the sides
>usually do not repeat the same openings but it may be important for the nunn
>match.
>
>I do not know if the difference in the results is because of learning and a
>difference between 15:5 for fritz and 11:9 for crafty seems to be too big to be
>explained by learning(It may be also an autoplayer problem).
>
>Uri
hi Uri,
if you play lilgoliath under chessbase gui you will see it has its usual
learning file.......just like it is under winboard. but for crafty it is not the
same it does not have any learinig file at all. I know that learning is a
important thing in nunn test, but it goes for both the engines not for crafty
alone, and even if crafty plays better with learinig then fritz, doesnt mean
that you have to clear it just make it play so that it doesnt win because of its
"superior" learinig code. It just_doesnt_add_up!! What if fritz has a better
timing code does it mean you will let fritz play with 2 min and craftywith 2min
50 secs? No you wont. Its just simple that the post by christophe didnt make any
sense.
thanks for you thoughts
pavel ;0
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.