I also tried large bonuses and so far 0.20 works better than 0.10 by about 20 ELO after 2,500 games. I tried 0.40 and it was significantly weaker, so this can definitely be overdone. The final result is 131 ELO for the 0.20 value. I will probably try 0.15 just to see what happens. I prefer to use lower values if it is all the same and I know that 0.10 is weaker than 0.20.
I can't try but so many things, so I'm optimizing and testing for 2000 playouts, which is a reasonable level for a PDA device. It's plays very quickly of course on a PC. On CGOS, looking at the bayes page it is rated 1475 based on 162 games, which compares well to ControlBoy (which is really anchorman) which has attained a rating of 1538 based on 74,000 games. But ControlBoy does 5000 playouts and this does only 2000. I guess I should put that one up too at 5000. - Don On Fri, 2008-10-31 at 17:29 -0400, Don Dailey wrote: > The idea is trivial: When scoring the moves at move selection time, > give a penalty for self-atari moves. I am only considering > non-capturing self-atari moves. I have only tried 2 values, 0.05 > and > 0.10 and 0.10 is working the best.
signature.asc
Description: This is a digitally signed message part
_______________________________________________ computer-go mailing list [email protected] http://www.computer-go.org/mailman/listinfo/computer-go/
