Stunning progress in the last weeks,
maybe it is a bit too early,
but i believe here is an paradigm shift underway.
So how do the old school chess programmers intend to catch up with LC0?
--
Srdja
LC0 - how to catch up?
Moderators: hgm, Rebel, chrisw
-
- Posts: 2658
- Joined: Wed Mar 10, 2010 10:18 pm
- Location: Hamburg, Germany
- Full name: Srdja Matovic
-
- Posts: 5228
- Joined: Thu Mar 09, 2006 9:40 am
- Full name: Vincent Lejeune
-
- Posts: 10297
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: LC0 - how to catch up?
I do not see a progress in the last weeks but a regression.smatovic wrote:Stunning progress in the last weeks,
maybe it is a bit too early,
but i believe here is an paradigm shift underway.
So how do the old school chess programmers intend to catch up with LC0?
--
Srdja
-
- Posts: 2658
- Joined: Wed Mar 10, 2010 10:18 pm
- Location: Hamburg, Germany
- Full name: Srdja Matovic
-
- Posts: 2658
- Joined: Wed Mar 10, 2010 10:18 pm
- Location: Hamburg, Germany
- Full name: Srdja Matovic
Re: LC0 - how to catch up?
I mean the last weeks since the selfplay and training started.I do not see a progress in the last weeks but a regression.
--
Srdja
-
- Posts: 7220
- Joined: Mon May 27, 2013 10:31 am
Re: LC0 - how to catch up?
I don't even know whether development using zero method or alpha beta method cost more electricity. Probably the first I guess.
They told me in 1994 if you know no better algorithm then use neural network. Don't know if that is still valid.
They told me in 1994 if you know no better algorithm then use neural network. Don't know if that is still valid.
-
- Posts: 2658
- Joined: Wed Mar 10, 2010 10:18 pm
- Location: Hamburg, Germany
- Full name: Srdja Matovic
Re: LC0 - how to catch up?
They told me in 2008 the same for Monte-Carlo methods...They told me in 1994 if you know no better algorithm then use neural network. Don't know if that is still valid.
--
Srdja
-
- Posts: 598
- Joined: Sun Aug 04, 2013 1:19 pm
Re: LC0 - how to catch up?
That's bad for you if you can't see the progress, because I can see a big progress.Uri Blass wrote:I do not see a progress in the last weeks but a regression.smatovic wrote:Stunning progress in the last weeks,
maybe it is a bit too early,
but i believe here is an paradigm shift underway.
So how do the old school chess programmers intend to catch up with LC0?
--
Srdja
-
- Posts: 2488
- Joined: Tue Aug 30, 2016 8:19 pm
- Full name: Rasmus Althoff
Re: LC0 - how to catch up?
Three things:
1) Alpha0 had that stunning hardware optimised for NNs although frameworks for GPUs already existed - but they wouldn't cut it, that's why Google came up with TPUs.
2) Low hanging fruit. Progress becomes harder over time.
3) You can't actually debug an NN.
1) Alpha0 had that stunning hardware optimised for NNs although frameworks for GPUs already existed - but they wouldn't cut it, that's why Google came up with TPUs.
2) Low hanging fruit. Progress becomes harder over time.
3) You can't actually debug an NN.
-
- Posts: 10297
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: LC0 - how to catch up?
I read in the forum and look in the graph and in the last days I do not see a progress based on the reports.Hai wrote:That's bad for you if you can't see the progress, because I can see a big progress.Uri Blass wrote:I do not see a progress in the last weeks but a regression.smatovic wrote:Stunning progress in the last weeks,
maybe it is a bit too early,
but i believe here is an paradigm shift underway.
So how do the old school chess programmers intend to catch up with LC0?
--
Srdja
There was a progress in march but the graph shows no progress in the last days and it is not only the graph but also what people report here.
http://lczero.org/