Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hiarcs7.32's learning bug

Author: blass uri

Date: 23:24:11 07/02/00

Go up one level in this thread


On July 03, 2000 at 01:30:17, blass uri wrote:

<snipped>
>After the game I tried to give hiarcs the position before 43.Bxb6 and was
>surprised to find that it can see that 43.Bxb6 is losing at depth 9 and also can
>find 43.Qe8 in less than 4 minutes

I tested Hiarcs7.32 with selectivity=7 and was surprised to find that it can
find 43.Qe8 even faster.

Selectivity=7 also can see at the same time control that I tested the default
hiarcs after the game(1 hour/40 move+3 hours/40 moves that means 4 hours for
moves 42-80) that 42.axb4 is bad at depth 10(the main line begin with axb4 Bxb4
and 42.axb4 failed low at depth 10).

It seems that hiarcs is less selective with selectivity=7 because it can see at
the same depth things that the default cannot see.

I have 2 questions about hiarcs.

1)Is learning off better than learning on?
2)Is selectivity=6 or 7 better than selectivity=5?

Uri



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.