Soberango 0.10.0 realized.
Add check extension.
First tests:
79-60-61 vs. Soberango 0.09.8 in 40/4.
Webpage: http://www.luisbtenis.com.ar/Soberango/Soberango.html
Soberango 0.00.0 realized.
Moderators: hgm, Rebel, chrisw
-
- Posts: 464
- Joined: Sat Feb 28, 2015 4:37 pm
- Location: Argentina
-
- Posts: 464
- Joined: Sat Feb 28, 2015 4:37 pm
- Location: Argentina
Re: Soberango 0.00.0 realized.
Thanks! This comment encourage me.... but I´m not sure how it will be from now on!SzG wrote:It seems its progress is accelerating.Luis Babboni wrote:Soberango 0.10.0 realized.
Add check extension.
First tests:
79-60-61 vs. Soberango 0.09.8 in 40/4.
Webpage: http://www.luisbtenis.com.ar/Soberango/Soberango.html
I´m working trying to find a bug that made 4 crashes in 900 games but the next step is try to add some bitboard stuff trying to increment the speed that I know will be not easy thing and next week its begin the hard part of the year in my RL.
-
- Posts: 464
- Joined: Sat Feb 28, 2015 4:37 pm
- Location: Argentina
Re: Soberango 0.00.0 realized.
Soberango 0.10.2 realized.
Fixed a bug that made engine crash sometimes when have only one possible move and checkmate in few moves against it is not avoidable.
Around 1150 ELO points compared with engines in the 40/4 CCRL rank.
Webpage: http://www.luisbtenis.com.ar/Soberango/Soberango.html
Fixed a bug that made engine crash sometimes when have only one possible move and checkmate in few moves against it is not avoidable.
Around 1150 ELO points compared with engines in the 40/4 CCRL rank.
Webpage: http://www.luisbtenis.com.ar/Soberango/Soberango.html
-
- Posts: 893
- Joined: Mon Jan 15, 2007 11:23 am
- Location: Warsza
Re: Soberango 0.00.0 realized.
perhaps it's time to start adding piece/square tables - seems like easy Elo gain.
Pawel Koziol
http://www.pkoziol.cal24.pl/rodent/rodent.htm
http://www.pkoziol.cal24.pl/rodent/rodent.htm
-
- Posts: 464
- Joined: Sat Feb 28, 2015 4:37 pm
- Location: Argentina
Re: Soberango 0.00.0 realized.
Thanks Pawel for the suggestion!PK wrote:perhaps it's time to start adding piece/square tables - seems like easy Elo gain.
But my plan is to let evaluation for the end, having the most simply evaluation function allow me to see easier how fine or wrong each new searching feature works.
And I think that being evaluation features the additions that could gives more skill level improvement, I have fear that after that I could lost interesting in the improvement of less important (in terms of skill level) features.
... or may be is just a whim...
In anyway I´ll take it in account cause my planned next search features could be beyond my abilities.... I´ll try with bitboards and later Transposition Tables.... so is possible that being unsuccessfull in those complicated plans... in few months I´ll be in piece/squares tables!!
-
- Posts: 464
- Joined: Sat Feb 28, 2015 4:37 pm
- Location: Argentina
Re: Soberango 0.00.0 realized.
Soberango 0.10.3 realized.
Now Soberango do not consider underpromotions in QS.
Not noticeable skill level change.
Changed next objetive, now is to add nullmove.
Website:
http://www.luisbtenis.com.ar/Soberango/Soberango.html
Now Soberango do not consider underpromotions in QS.
Not noticeable skill level change.
Changed next objetive, now is to add nullmove.
Website:
http://www.luisbtenis.com.ar/Soberango/Soberango.html
-
- Posts: 759
- Joined: Fri Jan 04, 2013 4:55 pm
- Location: Nice
Re: Soberango 0.00.0 realized.
What knowledge do Soberango have In his eval actually?
I share the Pawel advice : you can add some more knowledges in your engine because now it is more than 1000 elo .
Try to add psqt of course but a small bonus for passers for exemple (row dependant from 0 to +30 for exemple)
I think you can't see exactly the improvment of your search with an evaluation too simple ( material only)
It's just my advice , I'm not an expert
I share the Pawel advice : you can add some more knowledges in your engine because now it is more than 1000 elo .
Try to add psqt of course but a small bonus for passers for exemple (row dependant from 0 to +30 for exemple)
I think you can't see exactly the improvment of your search with an evaluation too simple ( material only)
It's just my advice , I'm not an expert
Isa download :
-
- Posts: 464
- Joined: Sat Feb 28, 2015 4:37 pm
- Location: Argentina
Re: Soberango 0.00.0 realized.
Thanks Daniel for your comment!
Not sure if I reach that point yet. I think I still have place to add some few more search features and then yes, I´ll must start with evaluation features.
Thanks again for your advice Daniel!
Just material.Daniel Anulliero wrote:What knowledge do Soberango have In his eval actually?
I use to think the opposite, at least for basic search features, but going little by little to most refined search features, there must be a point where the things turn to what you and Pawel and Sven say.Daniel Anulliero wrote: ...
I think you can't see exactly the improvment of your search with an evaluation too simple ( material only)
...
Not sure if I reach that point yet. I think I still have place to add some few more search features and then yes, I´ll must start with evaluation features.
Thanks again for your advice Daniel!
-
- Posts: 464
- Joined: Sat Feb 28, 2015 4:37 pm
- Location: Argentina
Re: Soberango 0.00.0 realized.
New URL for Soberango engine:
http://www.luisb.com.ar/Soberango/Soberango.html
http://www.luisb.com.ar/Soberango/Soberango.html
-
- Posts: 464
- Joined: Sat Feb 28, 2015 4:37 pm
- Location: Argentina
Re: Soberango 0.00.0 realized.
Soberango 0.12.0 realized.
Added material ratio to evaluation.
Web:
http://www.luisb.com.ar/Soberango/Soberango.html
Added material ratio to evaluation.
Web:
http://www.luisb.com.ar/Soberango/Soberango.html