Page 2 of 8
Re: A dangerous combination
Posted: Fri Mar 30, 2007 5:34 pm
by Dr.Wael Deeb
Tony Thomas wrote:Dr.Wael Deeb wrote:So with every new opponent you actually set the learning book function to 0,right?
Yes sir, so that means your book is actually pretty darn good. The scores are more or less learning independent.
Aha,now that was my real intention by asking,to hear these words from your golden mouth
Re: A dangerous combination
Posted: Fri Mar 30, 2007 5:37 pm
by Tony Thomas
Dr.Wael Deeb wrote:Tony Thomas wrote:Dr.Wael Deeb wrote:So with every new opponent you actually set the learning book function to 0,right?
Yes sir, so that means your book is actually pretty darn good. The scores are more or less learning independent.
Aha,now that was my real intention by asking,to hear these words from your golden mouth
Or it could be that Romi improved a lot.
Re: A dangerous combination
Posted: Fri Mar 30, 2007 5:41 pm
by Dr.Wael Deeb
It could be of course,but even with a remarkable improvement,it won't show this performance using a stuiped opening book....believe me!
Re: A dangerous combination
Posted: Fri Mar 30, 2007 5:45 pm
by Michael Sherwin
Tony Thomas wrote:Dr.Wael Deeb wrote:So with every new opponent you actually set the learning book function to 0,right?
Yes sir, so that means your book is actually pretty darn good. The scores are more or less learning independent.
With Grand07.pgn merged with the WD book as an extension, Romi performs 8% better! Though that results in a 194 MB file. I could send to you the DWD pgn files so you can make this huge book yourself as it seems better to first create the type 0 book and then merge the type 1 WD book. But, I need Dr. Wael Deeb's permission to send them to you.
Re: A dangerous combination
Posted: Fri Mar 30, 2007 5:47 pm
by Tony Thomas
Dr.Wael Deeb wrote:It could be of course,but even with a remarkable improvement,it won't show this performance using a stuiped opening book....believe me!
True, I havent seen Romi coming off the opening book with a negetive score. Then again out of the 12 opponents Romi plays against, only Arasan got a really good book. Yes, Francesca, Zappa, etc got a book, but I dont think the authors worked on the book as much as Jon Dart.
Re: A dangerous combination
Posted: Fri Mar 30, 2007 5:51 pm
by Dr.Wael Deeb
Michael Sherwin wrote:Tony Thomas wrote:Dr.Wael Deeb wrote:So with every new opponent you actually set the learning book function to 0,right?
Yes sir, so that means your book is actually pretty darn good. The scores are more or less learning independent.
With Grand07.pgn merged with the WD book as an extension, Romi performs 8% better! Though that results in a 194 MB file. I could send to you the DWD pgn files so you can make this huge book yourself as it seems better to first create the type 0 book and then merge the type 1 WD book. But, I need Dr. Wael Deeb's permission to send them to you.
Hi Michael,
no problems for sending the pgn files to Tony,he's my soul brother since the creation of mother earth
Re: A dangerous combination
Posted: Fri Mar 30, 2007 5:52 pm
by Michael Sherwin
They will be on the way shortly!
And THANKS!
Re: A dangerous combination
Posted: Fri Mar 30, 2007 5:54 pm
by Dr.Wael Deeb
Tony Thomas wrote:Dr.Wael Deeb wrote:It could be of course,but even with a remarkable improvement,it won't show this performance using a stuiped opening book....believe me!
True, I havent seen Romi coming off the opening book with a negetive score. Then again out of the 12 opponents Romi plays against, only Arasan got a really good book. Yes, Francesca, Zappa, etc got a book, but I dont think the authors worked on the book as much as Jon Dart.
As I wrote before,when I first read about the opening book learning function developed by Michael,I was so excited that I began working over an opening book for RomiChess waiting for the author to seek help creating a decent opening book
Cheers,
Dr.Wael Deeb
Re: A dangerous combination
Posted: Fri Mar 30, 2007 9:02 pm
by Tony Thomas
Dr.Wael Deeb wrote:Michael Sherwin wrote:Tony Thomas wrote:Dr.Wael Deeb wrote:So with every new opponent you actually set the learning book function to 0,right?
Yes sir, so that means your book is actually pretty darn good. The scores are more or less learning independent.
With Grand07.pgn merged with the WD book as an extension, Romi performs 8% better! Though that results in a 194 MB file. I could send to you the DWD pgn files so you can make this huge book yourself as it seems better to first create the type 0 book and then merge the type 1 WD book. But, I need Dr. Wael Deeb's permission to send them to you.
Hi Michael,
no problems for sending the pgn files to Tony,he's my soul brother since the creation of mother earth
I have been pretending to be your friend to get all your PGN files, now that I have them, what do I say
Re: A dangerous combination
Posted: Fri Mar 30, 2007 9:05 pm
by Tony Thomas
Michael Sherwin wrote:Tony Thomas wrote:Dr.Wael Deeb wrote:So with every new opponent you actually set the learning book function to 0,right?
Yes sir, so that means your book is actually pretty darn good. The scores are more or less learning independent.
With Grand07.pgn merged with the WD book as an extension, Romi performs 8% better! Though that results in a 194 MB file. I could send to you the DWD pgn files so you can make this huge book yourself as it seems better to first create the type 0 book and then merge the type 1 WD book. But, I need Dr. Wael Deeb's permission to send them to you.
I guess I will download Grand07.pgn tonight and then use that to test future versions of Romi. I will let the current version continue with the current file. Also, you know that changing the books a lot can skew the rating a lot, you wnt be able to tell weather it was the engine or the book unless I test the engine with both books.