LCZero Elo

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

How high will LCZero go in Elo in the next 12 months.

Poll ended at Fri Mar 30, 2018 10:12 pm

2000 or less.
6
15%
Between 2001 and 3400.
26
67%
Over 3400
3
8%
No guess. I dont understand the project.
4
10%
 
Total votes: 39

Dann Corbit
Posts: 12538
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: LCZero Elo

Post by Dann Corbit »

lucasart wrote:
CheckersGuy wrote:
lucasart wrote:For the moment, it's only ~1500 elo CCRL, and progress is starting to stall.

On the other hand, 12 months is a long time, and immense amount of resources are thrown at it. So perhaps 2000 is possible. But 3400, definitely not...
Lc0 is beating TSCP now so I would say it's clearly above 1500
No it doesnt.

See here:
http://talkchess.com/forum/viewtopic.php?t=66945

It gets smashed by Zurichess rated 1800...

Even I can still beat LC0 (though I do lose more games than I win I must admit).
This must be wrong then:
https://docs.google.com/spreadsheets/d/ ... edit#gid=0

They calculate 1806 Elo relative to stockfish variants of known strength.
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.
CheckersGuy
Posts: 273
Joined: Wed Aug 24, 2016 9:49 pm

Re: LCZero Elo

Post by CheckersGuy »

lucasart wrote:
CheckersGuy wrote:
lucasart wrote:For the moment, it's only ~1500 elo CCRL, and progress is starting to stall.

On the other hand, 12 months is a long time, and immense amount of resources are thrown at it. So perhaps 2000 is possible. But 3400, definitely not...
Lc0 is beating TSCP now so I would say it's clearly above 1500
No it doesnt.

See here:
http://talkchess.com/forum/viewtopic.php?t=66945

It gets smashed by Zurichess rated 1800...

Even I can still beat LC0 (though I do lose more games than I win I must admit).
Which version did you use ? The new v4 client with the latest net ?
The post is from the 28.3 which still used the old client. There was a major bug that was now fixed.
MonteCarlo
Posts: 188
Joined: Sun Dec 25, 2016 4:59 pm

Re: LCZero Elo

Post by MonteCarlo »

As Kai pointed out later in that post, his GPU is getting very, very low rollouts per second. The CPU version of Leela was much faster on his machine, and only performed about -119 rating to Zurichess.

Also, that was with network id 44, which is already substantially weaker than the strongest current one :)
CheckersGuy
Posts: 273
Joined: Wed Aug 24, 2016 9:49 pm

Re: LCZero Elo

Post by CheckersGuy »

The new net + the new client was about 300 selfplay elo stronger than id 44. Since some ppl were still using the v3 and not the v4 client the elo graph is completly messed up and the elo difference of 55 vs 44. is much lower in the graph.
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: LCZero Elo

Post by MikeB »

The second choice was a little too wide of a range ...400 points would have been a nice interval because a 400 point difference is significant...
elpapa
Posts: 211
Joined: Sun Jan 18, 2009 11:27 pm
Location: Sweden
Full name: Patrik Karlsson

Re: LCZero Elo

Post by elpapa »

Dann Corbit wrote:Against a pool of random movers
A pool of them? What possible difference could there be between one random mover and another?
BrendanJNorman
Posts: 2526
Joined: Mon Feb 08, 2016 12:43 am
Full name: Brendan J Norman

Re: LCZero Elo

Post by BrendanJNorman »

elpapa wrote:
Dann Corbit wrote:Against a pool of random movers
A pool of them? What possible difference could there be between one random mover and another?
Names :lol:
User avatar
CMCanavessi
Posts: 1142
Joined: Thu Dec 28, 2017 4:06 pm
Location: Argentina

Re: LCZero Elo

Post by CMCanavessi »

lucasart wrote:For the moment, it's only ~1500 elo CCRL, and progress is starting to stall.

On the other hand, 12 months is a long time, and immense amount of resources are thrown at it. So perhaps 2000 is possible. But 3400, definitely not...
Stall?

Code: Select all

   # PLAYER                                 :  RATING  PLAYED    W    D    L   (%)  D(%)  OppAvg  OppN  OppDiv
 184 Leela Chess Zero Gen 33 (ID 55) x64    :  1632.2     139   92    9   38    69     6  1441.8    25    24.9
 200 Leela Chess Zero Gen 30 (ID 44) x64    :  1506.3     200  106   22   72    59    11  1427.0    25    25.0
 208 Leela Chess Zero Gen 24 x64            :  1401.3     200   84   22   94    48    11  1427.0    25    25.0
 213 Leela Chess Zero Gen 20 x64            :  1317.8     200   64   27  109    39    14  1427.0    25    25.0
 226 Leela Chess Zero Gen 16 x64            :  1229.3     200   47   26  127    30    13  1427.0    25    25.0
 228 Leela Chess Zero Gen 14 x64            :  1185.3     200   42   20  138    26    10  1427.0    25    25.0
 232 Leela Chess Zero Gen 12 x64            :  1094.2     250   64   26  160    31    10  1278.6    49    40.8
 244 Leela Chess Zero Gen 10 x64            :   860.1      92   53   11   28    64    12   654.5    23    23.0
 247 Leela Chess Zero Gen 8 x64             :   791.6      92   45   17   30    58    18   654.5    23    23.0
 252 Leela Chess Zero Gen 6 x64             :   597.4      92   31   18   43    43    20   654.5    23    23.0
 256 Leela Chess Zero Gen 4 x64             :   369.0     150   43   18   89    35    12   623.1    15    15.0
Follow my tournament and some Leela gauntlets live at http://twitch.tv/ccls
Werewolf
Posts: 1795
Joined: Thu Sep 18, 2008 10:24 pm

Re: LCZero Elo

Post by Werewolf »

Carlos your progress is remarkable, keep going
User avatar
lucasart
Posts: 3232
Joined: Mon May 31, 2010 1:29 pm
Full name: lucasart

Re: LCZero Elo

Post by lucasart »

CheckersGuy wrote:The new net + the new client was about 300 selfplay elo stronger than id 44. Since some ppl were still using the v3 and not the v4 client the elo graph is completly messed up and the elo difference of 55 vs 44. is much lower in the graph.
What does the client version have to do with ELO ?
Theory and practice sometimes clash. And when that happens, theory loses. Every single time.