Krzysztof Grzelak wrote: ↑Tue Dec 20, 2022 12:30 pm
Your tests cannot be taken seriously.
Krzysztof, can you explain the reasoning behind your saying ?
Może odpowiem na Twoje pytanie Dariuszu po polsku a potem po angielsku. Dlaczego jego testów nie można brać poważnie - z powodu jednego bardzo ważnego czynnika. A tym czynnikiem jest książka debiutowa. Jego książka zła i nie nadaję się do testów. Aby nie być gołosłowny i kłamliwy, na pewnej stronie przeprowadzono test książek debiutowych, mówię tu o stronie https://sites.google.com/site/computers ... testings-2 i zauważ jak jego książka wypada - najgorzej z wszystkich książek. Jak to widzę to tylko przychodzi myśl w głowy - tragedia i rozpacz.
Why his tests cannot be taken seriously - because of one very important factor. His book is bad and I'm not up for testing. In order not to be groundless and deceitful, a test of debut books was conducted on a certain website, I'm talking about the site https://sites.google.com/site/computers ... testings-2 and notice how his book fares - the worst of all books. When I see it, the only thing that comes to mind is tragedy and despair.
Krzysztof Grzelak wrote: ↑Tue Dec 20, 2022 12:30 pm
Your tests cannot be taken seriously.
Krzysztof, can you explain the reasoning behind your saying ?
Why his tests cannot be taken seriously - because of one very important factor. His book is bad and I'm not up for testing. In order not to be groundless and deceitful, a test of debut books was conducted on a certain website, I'm talking about the site https://sites.google.com/site/computers ... testings-2 and notice how his book fares - the worst of all books. When I see it, the only thing that comes to mind is tragedy and despair.
Ouch! 100 game tests? The error margins are bigger than the elo diff in most cases. And the elo diff Cfish vs Kayra should be 15 elo a priori? Shouldn’t that be determined through extensive tests with some 30k games?
UHO accentuates the elo diff, so if with balanced openings we have 15 elo, with uho we should have 30 elo.
That page cannot be taken seriously.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
Krzysztof Grzelak wrote: ↑Tue Dec 20, 2022 12:30 pm
Your tests cannot be taken seriously.
Krzysztof, can you explain the reasoning behind your saying ?
Może odpowiem na Twoje pytanie Dariuszu po polsku a potem po angielsku. Dlaczego jego testów nie można brać poważnie - z powodu jednego bardzo ważnego czynnika. A tym czynnikiem jest książka debiutowa. Jego książka zła i nie nadaję się do testów. Aby nie być gołosłowny i kłamliwy, na pewnej stronie przeprowadzono test książek debiutowych, mówię tu o stronie https://sites.google.com/site/computers ... testings-2 i zauważ jak jego książka wypada - najgorzej z wszystkich książek. Jak to widzę to tylko przychodzi myśl w głowy - tragedia i rozpacz.
Why his tests cannot be taken seriously - because of one very important factor. His book is bad and I'm not up for testing. In order not to be groundless and deceitful, a test of debut books was conducted on a certain website, I'm talking about the site https://sites.google.com/site/computers ... testings-2 and notice how his book fares - the worst of all books. When I see it, the only thing that comes to mind is tragedy and despair.
If you test several books between 2 engines giving the same book to both, results will be meaningless. To test books you have to use the same engine against other ones (or even against itself) and give them different books.
dkappe wrote: ↑Tue Dec 20, 2022 4:00 pm
Ouch! 100 game tests? The error margins are bigger than the elo diff in most cases. And the elo diff Cfish vs Kayra should be 15 elo a priori? Shouldn’t that be determined through extensive tests with some 30k games?
UHO accentuates the elo diff, so if with balanced openings we have 15 elo, with uho we should have 30 elo.
That page cannot be taken seriously.
With more games, it can be even more of a tragedy.
A pleasant surprise. Given how difficult is to make progress, the fact that v3.2 clearly improves on a difficult test suite is impressive.
Not only did v3.2 find two more solutions, the time-to-solution was improved basically across the suite. Congrats to Komodo team!
Krzysztof Grzelak wrote: ↑Tue Dec 20, 2022 12:30 pm
Your tests cannot be taken seriously.
Krzysztof, can you explain the reasoning behind your saying ?
Może odpowiem na Twoje pytanie Dariuszu po polsku a potem po angielsku. Dlaczego jego testów nie można brać poważnie - z powodu jednego bardzo ważnego czynnika. A tym czynnikiem jest książka debiutowa. Jego książka zła i nie nadaję się do testów. Aby nie być gołosłowny i kłamliwy, na pewnej stronie przeprowadzono test książek debiutowych, mówię tu o stronie https://sites.google.com/site/computers ... testings-2 i zauważ jak jego książka wypada - najgorzej z wszystkich książek. Jak to widzę to tylko przychodzi myśl w głowy - tragedia i rozpacz.
Why his tests cannot be taken seriously - because of one very important factor. His book is bad and I'm not up for testing. In order not to be groundless and deceitful, a test of debut books was conducted on a certain website, I'm talking about the site https://sites.google.com/site/computers ... testings-2 and notice how his book fares - the worst of all books. When I see it, the only thing that comes to mind is tragedy and despair.
UHO book looks the best actually. It has the highest resolution. Do you understand what the purpose of an opening book for testing is?
dangi12012 wrote:No one wants to touch anything you have posted. That proves you now have negative reputations since everyone knows already you are a forum troll.
Maybe you copied your stockfish commits from someone else too?
I will look into that.
Krzysztof Grzelak wrote: ↑Tue Dec 20, 2022 12:30 pm
Your tests cannot be taken seriously.
Krzysztof, can you explain the reasoning behind your saying ?
Może odpowiem na Twoje pytanie Dariuszu po polsku a potem po angielsku. Dlaczego jego testów nie można brać poważnie - z powodu jednego bardzo ważnego czynnika. A tym czynnikiem jest książka debiutowa. Jego książka zła i nie nadaję się do testów. Aby nie być gołosłowny i kłamliwy, na pewnej stronie przeprowadzono test książek debiutowych, mówię tu o stronie https://sites.google.com/site/computers ... testings-2 i zauważ jak jego książka wypada - najgorzej z wszystkich książek. Jak to widzę to tylko przychodzi myśl w głowy - tragedia i rozpacz.
Why his tests cannot be taken seriously - because of one very important factor. His book is bad and I'm not up for testing. In order not to be groundless and deceitful, a test of debut books was conducted on a certain website, I'm talking about the site https://sites.google.com/site/computers ... testings-2 and notice how his book fares - the worst of all books. When I see it, the only thing that comes to mind is tragedy and despair.
UHO is not meant to be a "good" or "strong" book, for that, you should use either no book at all or a book like "Perfect 2021".
UHO is meant to be a "bad" or "biased" book to exaggerate engine strength differences and realize more easily and with less games if your engine is superior to the competition, not to be used by humans.
Dragon 3.2 (C) 2022 Don Dailey, Larry Kaufman, Mark Lefler, Dmitry Pervov, and Dietrich Kappe
using hardware POPCNT
info string Licensed to illegal copy
info string embedded NN is loaded
"Sounds' like you got an illegal copy (?). Did you actually download it from the Komodo/Dragon website?
Dragon 3.2 (C) 2022 Don Dailey, Larry Kaufman, Mark Lefler, Dmitry Pervov, and Dietrich Kappe
using hardware POPCNT
info string Licensed to illegal copy
info string embedded NN is loaded
"Sounds' like you got an illegal copy (?). Did you actually download it from the Komodo/Dragon website?
Downloaded from original website as a customer!!
True...why would an 'illegal' copy actually say 'illegal'. Perhaps it's some internal thing to ward of pirates or people who share their copies.
Sopel wrote: ↑Tue Dec 20, 2022 5:19 pm
UHO book looks the best actually. It has the highest resolution. Do you understand what the purpose of an opening book for testing is?
