SheikhYerbouti wrote: ↑Tue Feb 04, 2020 3:17 am
So if Leela's relatively bad playing from g4 what happens with SF self-play from ply 2? More likely a draw?
Exactly, when matched Stockfish can't beat itself from the black side, that's the main reason I thought I could draw with 1.g4. Even on my losing positions against Havey, Stockfish wasn't being able to win from the black side (and that's why I took so long to resign.)
But when you guide Stockfish as black against any white defense, it's a very different story, now it's Stockfish as white that can't defend the position. And when you try to guide its defense as white, it seems useless, everything seems losing, so far, at least for 2.g5 and 2.Bg2, and they looked better than the alternatives (so much that instead of trying to defend the white side with another move, I switched to attacking white's last line of defense.)
Is 1.g4 lost??
I think beat itself may be dependent on time control.
I wonder if stockfish really cannot beat itself with black after 1.g4 when it get a good hardware and 24 hours for every move.
Uri Blass wrote: ↑Tue Feb 04, 2020 5:45 am
I think beat itself may be dependent on time control.
I wonder if stockfish really cannot beat itself with black after 1.g4 when it get a good hardware and 24 hours for every move.
If that happens then you can come 5 years later and claim the feat wasn't done with "good hardware" anymore.
[d]3qr1k1/p4pp1/1rpbbnn1/3p2Bp/3P3P/1PNQPN2/P4PB1/R4K1R b - - 2 17
No need to delay, other moves get down to around -2.5 according to SF11 and LC0 doesn't like them either.
Zenmastur wrote: ↑Tue Feb 04, 2020 3:52 amI had 15.Na4 at -1.71 and this was the best move.
For the record, I had 15.Na4 also as best and this is what I was expecting.
LC0's scores and deeper analysis made a difference here. I'd have to run the analysis again to show you exactly what happened and the lines are still valid for the game so it will have to wait a bit.
I'm OK with waiting. Just for reference what hardware are you using? I don't think I've seen you post it.
Regards,
Zenmastur
Only 2 defining forces have ever offered to die for you.....Jesus Christ and the American Soldier. One died for your soul, the other for your freedom.
jp wrote: ↑Tue Feb 04, 2020 3:01 am
Can you keep some examples of these lines Leela doesn't see but agrees change everything when you feed them into it and post afterwards, so we can look at them?
Just match Leela v Stockfish and see for yourself.
I don't have the hardware (for Lc or SF) that would allow a meaningful comparison.
Okay, but then the difference between non-spoonfed Leela (-1.3) and spoonfed Leela (say -2.0, though you actually said -1.6) is "only" 0.7 (or as little as 0.3). I'm not convinced that this is then evidence that Leela's problems are "tactical" here.
Except that ALL the variations I've seen on this game where Leela showed a score of -1.60 or worse have been COMPLETELY LOST for white, so it could as well be showing -9.99, because those are positions where black has string of moves that mate in X.
That just indicates Leela on whatever hardware you run it on is weak. It does not show that its problems or these positions are "tactical". Part of the issue is of course agreeing what "tactical" means. I've already said "strategy vs. tactics" is not a meaningful distinction for engines, because all they do is calculate. But we could apply judgements about what humans regard as "tactical", etc. If Leela shows only -2.0 and it's really mate in 30 for Black, but there were no blunders by human standards, then that's hardly evidence that this is a "tactical" issue. I mean, you spoonfeed Leela the winning line, when it still only goes to -2.0, so in the spoonfed case it's not "overlooking a tactic". It just cannot evaluate the line or position properly. Normally, if you spoonfeed something or someone a winning line and it/he still doesn't know what's going on, you say it has no understanding, not that it's "bad at tactics".
Isn't selfplay the best way to evaluate position!? SF 11 games at tournament level from move 18. After 40 plies 1. try -2,59 and 2. try -3,64. I dare to say, that black has advantage .
Jouni wrote: ↑Tue Feb 04, 2020 3:39 pm
Isn't selfplay the best way to evaluate position!? SF 11 games at tournament level from move 18. After 40 plies 1. try -2,59 and 2. try -3,64. I dare to say, that black has advantage .
What does "tournament level" mean? What average search depth, for example?
.
.