You have 75% probability of the closed openings using this book.
The test set you used in the first match had it 50-50
What engine do you think Salvo used to create and tune the book
NAUM 4.2
Moderator: Ras
-
naum
Re: NAUM 4.2
This is what I'm saying.
You have 75% probability of the closed openings using this book.
The test set you used in the first match had it 50-50
What engine do you think Salvo used to create and tune the book
You have 75% probability of the closed openings using this book.
The test set you used in the first match had it 50-50
What engine do you think Salvo used to create and tune the book
-
naum
Re: NAUM 4.2
I absolutely agree. This position is on my to do list for quite a while, since I see it often.
Problem is the limited time I have which I dedicate to things that will improve elo (fixing this would most likely have no effect on the elo).
But this is a very common position and should be fixed, so I will try to do it in the next version (together with rook and pawn fortress against the queen which is also on my to do list after I saw Stockfish use it against Naum
Problem is the limited time I have which I dedicate to things that will improve elo (fixing this would most likely have no effect on the elo).
But this is a very common position and should be fixed, so I will try to do it in the next version (together with rook and pawn fortress against the queen which is also on my to do list after I saw Stockfish use it against Naum
-
Uri Blass
- Posts: 10896
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: NAUM 4.2
I do not believe that evaluation knowledge cannot improve the elo.
If Naum choose to go to drawn endgame with evaluation of +3 pawns and not to winning position with evaluation of +2 pawns then it is going to lose 1/2 point.
Uri
If Naum choose to go to drawn endgame with evaluation of +3 pawns and not to winning position with evaluation of +2 pawns then it is going to lose 1/2 point.
Uri
-
naum
Re: NAUM 4.2
I know it's not logical. but it's true.
Knowledge of this position being drawn will not help, because most of the positions leading to it are also drawn. So yes, Naum will avoid this position but it will end up in another drawn position causing 0 elo effect.
In my experience knowledge like this has to be tested very well, because it often hurts engine in games. I actually had to remove some of my drawn endgame knowledge from search and eval because engine played worst with it.
That's a main problem. It's easy to implement, but it takes a lot of time to test and make sure it doesn't hurt the performance. And I have a very little time...
Knowledge of this position being drawn will not help, because most of the positions leading to it are also drawn. So yes, Naum will avoid this position but it will end up in another drawn position causing 0 elo effect.
In my experience knowledge like this has to be tested very well, because it often hurts engine in games. I actually had to remove some of my drawn endgame knowledge from search and eval because engine played worst with it.
That's a main problem. It's easy to implement, but it takes a lot of time to test and make sure it doesn't hurt the performance. And I have a very little time...