Le Mardi 7 Février 2006 01:28, alain Baeckeroot a écrit : > it seems less clear that my patch does something good, but i m going on... >
Hi i m doing systematical investigation of the choice of the distance and bonus decrease in my patch. here bonus = maximum of all ( opponent val / (max_dist +1) ) i ve plotted cumulative unexpected pass(+1) / fail (-1) versus game_status for 19x19 tests, because i have nearly 400 breakage. it seems that it can do something good for game status between 0.5 and 0.8. i ll open a ticket when i get something clean. i dont remember how to get a view of the board, with the tests results, it would ease this huge analyse :) Any comment/suggestion/critic is welcome Alain
19-regress-float-1divD.png
Description: PNG image
_______________________________________________ gnugo-devel mailing list gnugo-devel@gnu.org http://lists.gnu.org/mailman/listinfo/gnugo-devel
