Ya, still it would be nice if they made it a usable engine. But their probably not going to do that, right?
What is the best opening move according to engines?
Sort:
Oct 8, 2020
0
#83
llama wrote:
stassneyking wrote:
you certainly can't just say a computer is better than Alpha Zero without it beating it in a match.
Umm, yes you can. Everyone knows AZ is not the strongest. It's easy to know because AZ and stockfish played a match so that benchmarked AZ, and then stockfish improved in later versions as benchmarked by playing other engines.
Then you just compare how much AZ was better to how much SF improved. Ez
This is not a logically sound argument. The games would have been totally different.
Oct 8, 2020
0
#85
emchel wrote:
Ya, still it would be nice if they made it a usable engine. But their probably not going to do that, right?
It ran on specialized hardware that costs more than your house... so no, they're "not going to do that" lol 
Leela combined enthusiast's computing power to train it. It's both better than alpha zero was, and is is available to normal people.
And now they even have the hybrid stuff with stockfish nnue... so yeah, AZ is a thing of the past.
Oct 8, 2020
0
#86
stassneyking wrote:
It's not some kind of linear equation.
You're right, the rating formula has variables as exponents... lol
stassneyking wrote:
Why would it matter if it's in the DeepMind lab? If it outplays stockfish in a match it wins.
Because a match where two engines are both being set up and run by the team members who created one of those engines, and then reporting results from behind closed doors, is not something you would want to officially trust. A neutral 3rd party needs to run a match for Alpha Zero to be considered "champion". Alpha Zero not only is not a champion, it has never even competed. It's a great test engine in a lab.
Fact.
Oct 8, 2020
0
#88
Leela has definitely not been proven to be better than Alpha, and stockfish nnue is taking all kinds of losses so I don't really know what you're talking about.
Oct 8, 2020
0
#89
I don't suspect any kind of conspiracy, but I suppose you could be right.
Oct 8, 2020
0
#90
@stassneyking
For someone whose daily rating is 1000 points higher than his bullet rating you sure seem to not know much about engines.
Oct 8, 2020
0
#91
llama wrote:
emchel wrote:
Ya, still it would be nice if they made it a usable engine. But their probably not going to do that, right?
It ran on specialized hardware that costs more than your house... so no, they're "not going to do that" lol
Leela combined enthusiast's computing power to train it. It's both better than alpha zero was, and is is available to normal people.
And now they even have the hybrid stuff with stockfish nnue... so yeah, AZ is a thing of the past.
Ok yes, that was kind of a foolish question by me
Oct 8, 2020
0
#92
lol I hate bullet, classical specialist here. I must admit I don't know that much about engines, just enjoyed reading Game Changer.
Oct 8, 2020
0
#94
I mean... your rapid and blitz aren't so high either... and you're the 5th highest rated player on the whole site for daily.
Just sayin'
Oct 8, 2020
0
#95
It's all I've been playing lately and I've been working very hard at it. I still throw away wins constantly.
stassneyking the conclusion of this discussion is your statements have been a joke, period.
Even the posters who were trying to slant to your side are shocked by your cluelessness of the subject what a joke.
Oct 8, 2020
0
#99
lol no need for harsh words. I don't think you understand the argument that I'm making, but that's okay. I don't have all day to discuss this so I won't bother repeating myself. Thanks for sharing your thoughts on the subject.
I really would like to see for alphazero games, don't know why they don't make it play in engine vs engine tournaments.
Because it wasn't a chess engine... at least not only. It also beat the the best shogi and go engines... the point was never to play games, it was publicity for their research / engineering.