Author: Dann Corbit
Date: 10:29:27 02/02/06
Go up one level in this thread
On February 02, 2006 at 06:15:13, David B Weller wrote: >Hi Dann & Uri, > >I WAS mostly thinking of positional suites, but now that you mention it, the >same holds true for tactical ones... > >Namely, if, in the final analysis, the recap extension (eg) isnt really worth >its price, but many 'interesting' positions in tactical suites are solved when >using it, does this not give an inacurate etimate of its value? > >I dont think this is anything most have not already figured out. Thats why they >test with real games ... > >So the moral of the story is: Use test suites to discover gross error in eval or >search, but do not tune by them.... I think that everyone agrees to that. The biggest value of test suites is: 1. Learn about holes in your evaluation while you are early in developing your engine 2. Make sure that a recent change didn't knock a great big hole in something important in your chess engine
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.