FIBS Board backgammon forum

NiceRollerBot settings in gnubg?

Author Topic: NiceRollerBot settings in gnubg?  (Read 6595 times)

Offline eldadrosi

  • Newbie
  • *
  • Posts: 3
NiceRollerBot settings in gnubg?
« on: April 01, 2005, 06:20:42 PM »
this new robot is the best on fibs and plays fast.

i would very much want to know what settings does he use with the gnubg

program so i have the same excelent results offline.

...better then mgnuwclass and plays fast...     very good !!!  :D  :cool:


i hope to see the robot plays all the time  :)  

FIBS Board backgammon forum

NiceRollerBot settings in gnubg?
« on: April 01, 2005, 06:20:42 PM »

Offline adamosad

  • Fibsboarder plus
  • ****
  • Posts: 359
    • http://fibsleagammon.free.fr/Rating_Championship.htm
NiceRollerBot settings in gnubg?
« Reply #1 on: April 01, 2005, 07:29:41 PM »
Quote
i hope to see the robot plays all the time  :)
I agree completely with you!!! :yes:  

FIBS Board backgammon forum

NiceRollerBot settings in gnubg?
« Reply #1 on: April 01, 2005, 07:29:41 PM »

Offline MND

  • Fibsboarder
  • ***
  • Posts: 44
    • http://
NiceRollerBot settings in gnubg?
« Reply #2 on: April 02, 2005, 04:09:45 PM »
Nicerollerbot, parlorbot and mgnu_wclass all use almost the same settings, Gnu 2-ply supremo.
parlorbot use a ligth move filter, in order to speed things up, and wclass a larger one.

The difference in playing strength between these 3 should be 5 fibs rating points  or so.
The rest depends on luck, number of droppers, new players who start to play with the bots,  the sligthly inaccurate fibs formula etc etc.


 

Offline eldadrosi

  • Newbie
  • *
  • Posts: 3
NiceRollerBot settings in gnubg?
« Reply #3 on: April 02, 2005, 06:57:54 PM »
Quote
Nicerollerbot, parlorbot and mgnu_wclass all use almost the same settings, Gnu 2-ply supremo.
parlorbot use a ligth move filter, in order to speed things up, and wclass a larger one.

The difference in playing strength between these 3 should be 5 fibs rating points  or so.
The rest depends on luck, number of droppers, new players who start to play with the bots,  the sligthly inaccurate fibs formula etc etc.
i wish there were an average command where i can take the average rating of the last x games ...

supremo is gnu 2 play+large move filter - you say its mgnuwclass

parlorbot is gnu 2 play+small move filter - (not supremo)

nicerollerbot the maker now told me its gnu 2 play +normal move filter(not supremo)

also maybe because nrb plays now only 1 point games its better then the others.

maybe cube handling in gnu is not so good  :unsure:  

FIBS Board backgammon forum

NiceRollerBot settings in gnubg?
« Reply #3 on: April 02, 2005, 06:57:54 PM »

Offline adamosad

  • Fibsboarder plus
  • ****
  • Posts: 359
    • http://fibsleagammon.free.fr/Rating_Championship.htm
NiceRollerBot settings in gnubg?
« Reply #4 on: April 02, 2005, 09:00:17 PM »
I think you confuse a little. All use Gnu 2-ply. They could use for example Gnu 0-ply but i think this is weaker.

The difference is only the filter they use.....

Offline amarganth

  • Fibsboarder +
  • ***
  • Posts: 89
    • http://www.RedAndBlue.ch.vu
NiceRollerBot settings in gnubg?
« Reply #5 on: April 02, 2005, 11:32:28 PM »
Thanks for that nice compliment, eldadrosi

The settings of the mgnu bots is described here. For instance, mgnu_advanced has a "noise".

I think, NiceRollerBot has a higher rating than other bots at the moment, because I take care of the unfinished matches. I have some limits, and I check RepBot after an invitation.

NiceRollerBot currently has only 9 unfinished matches, the other bots have all around 80-130. I think, most of this games, the bots are winning. May be it's luck (I don't think so), may be that's the reason.

Greetings

amarganth
« Last Edit: April 02, 2005, 11:36:16 PM by amarganth »
To be is to do
          Sokrates
To do is to be
          Sartre
Do be do be do
          Sinatra

Offline eldadrosi

  • Newbie
  • *
  • Posts: 3
NiceRollerBot settings in gnubg?
« Reply #6 on: April 16, 2005, 08:37:14 PM »
i have a great idea for the bots.

you said you think the high rating is due to a few saved games.

i dont think so... !

dropper statistics i red see only a mild influence on rating over time
something between 5 to 20 points max(due to unfinished games).

its because it plays only 1 point.

i noticed(though i play only 1 point on fibs) that the few times i played against

mgnuwclass 3 points and more almost every time it doubled i should drop!

95 % right ! so its also because i have only 1700 and it has 2000 but still

computer doesnt make mistakes (i mean it doesnt bluff or try to frighten you)so i dropprd each time it doubled and the computer loses valuable points...

also in the last computer olympic the computer that played against mgnu won
because it moved its doubling mode to a more agressive one
 

Tags:
 

TinyPortal © 2005-2018