tmokonen wrote: ↑Fri Aug 24, 2018 6:16 am
So... the main server nets are gone, never to return?
The old server is still running under the url http://oldmain.lczero.org/
We'll have all networks conveniently available for download in one place eventually.
"it seems above 400,000 nodes ELO growth diminishes heavily (1-7 ELO per doubling NPM) "
"From my tests Leela is far stronger than Stockfish at very short time controls, however given longer thinking time and more nodes, SF outscales a 20 block Leela ID 480 "
I know that A0 scaled better than stockfish8 and certainly earned elo from more time and it seems that the lc0 team has a serious bug or do something clearly different than A0 or both of them.
tmokonen wrote: ↑Fri Aug 24, 2018 6:16 am
So... the main server nets are gone, never to return?
The old server is still running under the url http://oldmain.lczero.org/
We'll have all networks conveniently available for download in one place eventually.
Uri Blass wrote: ↑Fri Aug 24, 2018 7:44 am
I know that A0 scaled better than stockfish8 and certainly earned elo from more time and it seems that the lc0 team has a serious bug or do something clearly different than A0 or both of them.
A0 scaling vs SF from that crappy paper is totally bogus. It is a fact! I wonder when are ppl gonna stop citing that BS graph...
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
Uri Blass wrote: ↑Fri Aug 24, 2018 7:44 am
I know that A0 scaled better than stockfish8 and certainly earned elo from more time and it seems that the lc0 team has a serious bug or do something clearly different than A0 or both of them.
A0 scaling vs SF from that crappy paper is totally bogus. It is a fact! I wonder when are ppl gonna stop citing that BS graph...
I prefer to assume they are not lying but if they are lying then what is the reason for them not to tell the public that A0 is good at bullet time control.
It is better to be strong both at bullet time control and at long time control and not only at long time control.
Milos wrote: ↑Fri Aug 24, 2018 11:49 pm
A0 scaling vs SF from that crappy paper is totally bogus. It is a fact! I wonder when are ppl gonna stop citing that BS graph...
I prefer to assume they are not lying
Shouldn't assume anything. "Lying" is also an assumption about why it's wrong.
I think Milos is saying the SF line is totally wrong. SF devs or testers (maybe Pohl) should have a graph somewhere to see.
Uri Blass wrote: ↑Fri Aug 24, 2018 7:44 am
I know that A0 scaled better than stockfish8 and certainly earned elo from more time and it seems that the lc0 team has a serious bug or do something clearly different than A0 or both of them.
A0 scaling vs SF from that crappy paper is totally bogus. It is a fact! I wonder when are ppl gonna stop citing that BS graph...
I prefer to assume they are not lying but if they are lying then what is the reason for them not to tell the public that A0 is good at bullet time control.
It is better to be strong both at bullet time control and at long time control and not only at long time control.
I also think it's more plausible that we either have some sort of bug that hurts scaling, or simply haven't figured out some part of the AlphaZero approach properly yet. One experiment being tried now is changing the search parameters, i.e. PUCT, FPU reduction and futile move aversion constant. I'm confident that this will eventually be solved, it will just take some time now that this issue has come to light.