Author: Odd Gunnar Malin
Date: 11:15:44 08/17/02
Go up one level in this thread
On August 16, 2002 at 23:20:17, Will Singleton wrote: >On August 16, 2002 at 16:37:19, Peter Fendrich wrote: > >>On August 16, 2002 at 15:12:36, Odd Gunnar Malin wrote: >> >>>On August 16, 2002 at 13:14:37, Will Singleton wrote: >>> >>>>On August 16, 2002 at 03:09:05, Odd Gunnar Malin wrote: >>>> >>>>> >>>>>I made a tool for automatic tuning of an engine but there have been zero >>>>>feedback on this so I guess most found this sort of tools useless. >>>>> >>>>>Odd Gunnar >>>> >>>>What tool was this? >>> >>>This was a tool that looped through a testset (epd-file) and adjusted evaluation >>>for each run. >>> >>>If you f.ex. had command in the engine like this: >>> >>>set bishoppair 40 >>>set rookonseventhrank 30 >>>set doublepawnpenelty -20 >>>etc. >>> >>>You tell the program with a input file like this: >>> >>>(Command|Min value|Max value|Step) >>> >>>set bishoppair|0|80|10 >>>set rookonseventhrank|0|70|10 >>>set doublepawnpenelty|-50|0|10 >>> >>> >>>It could loop throug all possible values for these settings and save the best >>>setting to a file. >>> >>>Ex. first run >>> >>>set bishoppair 0 >>>set rookonseventhrank 0 >>>set doublepawnpenelty -50 >>> >>>Next run would be with step = 10 for bishoppair >>> >>>set bishoppair 10 >>>set rookonseventhrank 0 >>>set doublepawnpenelty -50 >>> >>>etc. >>> >>>If you download the file >>> >>>http://home.online.no/~malin/sjakk/download/GradualTest.zip >>> >>>It should be in there, it is called adjusteval.exe >>> >>>The source file is here: >>> >>>http://home.online.no/~malin/sjakk/download/GradualTest_src.zip >>> >>> >>>The hard task is naturly to find enough position where the right move could be >>>found with pure evaluation, or at least with a searchdept of 2 - 3. >>> >>>I used Dann Corbit's quiet possition for this with first take away all moves >>>that needed some searchdepth to find. I used the gradualtest (also in the >>>packed) for this sou I could automatic the process. >>> >>>I also had another thing in my thought when I made it. To make personalities for >>>my engine. It could be easier to find position for this because you don't look >>>for the overall best move. >>> >>>Odd Gunnar >> >>I tried that but couldn't get it to work for some reason so I build the function >>inside Terra instead.... I will check a little on the tool before my next release of my engine, I see that at least the readme file have to be rewrited. >> >>Peter > >Did either of you find a large enough test-suite to make this worthwhile? What >was your experience as to the benefits vs. hand-tuning? > >Will I'm not at home now so this is from half year old memory. I needed some automatic tuning before the CCT4 tournament. Because the little time to do this 'I made the adjusteval utility and for the position I took Dann's big epd file from GM-games and let Crafty run through it with search dept set to 1. Then I did a new run with the result file and set the search depth to 2 etc. up to 8 plies. Then I took Yace and did the same. All this run can be automated with GradualTest so you just set it up and let it run over night. The result file was with about 1000 position. I think Dann have it on his site but there is no ftp program on this PC to check. This isn't the best way to do it, but somehow you need a big base of position where all GM's agree with the best move from evaluation before you try to find position that you want because of the playing style you want. I have asked some web-shop etc. to search for positional testbooks where the search depth isn't to deep, but no one have found it yet. How my tuning was helping is hard to say, because my engine is so new that there is many changes in the code as well. It did do a jump of ca. 100 elo as best score on ICC, but I have not done an analyses of the result. Hand-tuning isn't easy either. As I said earlier in the thread, my use will be more to find personalities for the engine. I have big bases of games with local players where I will select some of these plyers to be guideline for the engines personalities. I will see when I get home if I get time to run some tests. If other's have time they can find all needed in the file: http://home.online.no/~malin/sjakk/download/CCT4-11.zip It should have both the engine and the resulting ini file. For my hand-tuned variables there is just to mask out all setting in the inifile the inifile. (Obs! hashsize is set very big so adjust it down) It will of course only compare automatic tuning with my hand-tuning. Others hand-tuning will probably give other results. :) Odd Gunnar
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.