Author: KarinsDad
Date: 12:44:28 07/27/99
Go up one level in this thread
On July 27, 1999 at 13:47:08, Will Singleton wrote: [snip] > >The bm/am thing is a little tough, since most amateur programs do not support >n-best moves. That's weird as that was one of the first things I wanted for my program. I organized the epd's a little differently, separating the am's >from the bm's, and it seems to work for me. A couple of comments: > >Pos0001, white should play b3 or b4, to get the outside passer. Mine likes >passer's, especially outside ones, but didn't want to play the right move. So I >tweaked up the bonuses and finally it plays b3. I have to be careful with this, >however, since the problem *might* have been that other terms were being >evaluated too high relative to the passed pawn, so they should have been >reduced, rather than the passed pawn increased. I can see where this would be a problem. I was setting up some default piece location delta scores (1/6th pawn down to 1/20th pawn) last night and couldn't decide whether to make it more important to have edge pawns take towards the center or to stay put so they can be pushed later. There are a lot of factors to consider and I'm positive that the positional elements will have a lot of seriously out of whack eval scores. > >Pos0000: I don't consider this the proper time to centralize the king, still too >many pieces on the board. Hiarcs 6 switches to Ncxe2 after a minute, and my >prog likes Ndxe2. > >Will
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.