Re: [Computer-go] A new ELF OpenGo bot and analysis of historical Go games
On 17/02/19 23:24, Hiroshi Yamashita wrote: > Hi Ingo, > >> * How strong is the new ELF bot in comparison with Leela-Zero? > > from CGOS BayesElo, new ELF(ELFv2) is about +100 stronger than Leela-Zero. We ran a test match and ELFv2 lost 34 - 62 against LZ-204 at 1600 visits each, so that's about +100 Elo in favor of LZ at visits rather than time parity. This would mean going 800p -> 400p gives -200 Elo? Seems more than I would expect. I think I'd want to do more testing before forming an opinion :-) > Leela Zero's playout is half. Because its net size is double. > http://www.yss-aya.com/cgos/19x19/bayes.html Smaller networks have more overhead (heads etc are fixed), so it's closer to ~60%. Probably depends on the GPU too. -- GCP ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
Re: [Computer-go] A new ELF OpenGo bot and analysis of historical Go games
Dear Hiroshi, thanks a lot for your explanation! Best regards, Ingo. > Gesendet: Sonntag, 17. Februar 2019 um 23:24 Uhr > Von: "Hiroshi Yamashita" > An: computer-go@computer-go.org > Betreff: Re: [Computer-go] A new ELF OpenGo bot and analysis of historical Go > games > > Hi Ingo, > > > * How strong is the new ELF bot in comparison with Leela-Zero? > > from CGOS BayesElo, new ELF(ELFv2) is about +100 stronger than Leela-Zero. > > Rating Network size(Resnet layers x filters) > LZ_05db_ELFv2_p800 355420x256 > LZ_d13c_ELFv1_p800 353020x224 second release ELF > LZ_62b541_ELF_p800 350020x224 first release ELF > LZ_204_05d1_p400343340x256 latest(2019-02-13) LZ > > p800 means 800 playout/move. > p400 means 800 playout/move. > Leela Zero's playout is half. Because its net size is double. > http://www.yss-aya.com/cgos/19x19/bayes.html > > Thanks, > Hiroshi Yamashita > > On 2019/02/17 1:29, "Ingo Althöfer" wrote: > > Hi Remi, > > thanks you for the link. > > > > A few questions (to all who know something): > > > > * How strong is the new ELF bot in comparison with Leela-Zero? > > > > * How were komi values taken into account when analysing old go games with > > help of ELF? > > > > * How often does ELF propose moves played by AlphaGo (for instance in the > > games > > with Fan Hui, Lee Sedol, and in the sixty games from December 2017)? > > > > * Does ELF understand that the strength of AlphaGo increased from October > > 2015 to May 2017? > > > > Cheers, Ingo. > > ___ > > Computer-go mailing list > > Computer-go@computer-go.org > > http://computer-go.org/mailman/listinfo/computer-go > > > ___ > Computer-go mailing list > Computer-go@computer-go.org > http://computer-go.org/mailman/listinfo/computer-go ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
Re: [Computer-go] A new ELF OpenGo bot and analysis of historical Go games
Hi Ingo, * How strong is the new ELF bot in comparison with Leela-Zero? from CGOS BayesElo, new ELF(ELFv2) is about +100 stronger than Leela-Zero. Rating Network size(Resnet layers x filters) LZ_05db_ELFv2_p800 355420x256 LZ_d13c_ELFv1_p800 353020x224 second release ELF LZ_62b541_ELF_p800 350020x224 first release ELF LZ_204_05d1_p400343340x256 latest(2019-02-13) LZ p800 means 800 playout/move. p400 means 800 playout/move. Leela Zero's playout is half. Because its net size is double. http://www.yss-aya.com/cgos/19x19/bayes.html Thanks, Hiroshi Yamashita On 2019/02/17 1:29, "Ingo Althöfer" wrote: Hi Remi, thanks you for the link. A few questions (to all who know something): * How strong is the new ELF bot in comparison with Leela-Zero? * How were komi values taken into account when analysing old go games with help of ELF? * How often does ELF propose moves played by AlphaGo (for instance in the games with Fan Hui, Lee Sedol, and in the sixty games from December 2017)? * Does ELF understand that the strength of AlphaGo increased from October 2015 to May 2017? Cheers, Ingo. ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
Re: [Computer-go] A new ELF OpenGo bot and analysis of historical Go games
"The sudden overall increase in agreement in 2016 also reinforces the belief that the introduction of powerful AI opponents has boosted the skills of professional players. That apparent correlation isn't conclusive — it's possible that humans have gotten markedly better for some other reason — but it's an example of how a system trained to carry out a given task can also provide wide-ranging analysis of a larger domain, both in the present and from a historical perspective. " They are using their go AI to measure the strength of human go players, and use its analysis to 'prove' that copying the moves of AIs has made humans stronger (or at least, quantify this increase in strength). There is a huge bias there. I think this is like asking a fisherman whether chefs who specialize in fish have better cooking skills than chefs who specialize in meat... 2019-02-16 17:49 UTC+01:00, J. van der Steen : > > And most important: > >* Does ELF know the meaning of life? > > On 16/02/2019 17:29, "Ingo Althöfer" wrote: >> Hi Remi, >> thanks you for the link. >> >> A few questions (to all who know something): >> >> * How strong is the new ELF bot in comparison with Leela-Zero? >> >> * How were komi values taken into account when analysing old go games with >> help of ELF? >> >> * How often does ELF propose moves played by AlphaGo (for instance in the >> games >> with Fan Hui, Lee Sedol, and in the sixty games from December 2017)? >> >> * Does ELF understand that the strength of AlphaGo increased from October >> 2015 to May 2017? >> >> Cheers, Ingo. >> ___ >> Computer-go mailing list >> Computer-go@computer-go.org >> http://computer-go.org/mailman/listinfo/computer-go >> > ___ > Computer-go mailing list > Computer-go@computer-go.org > http://computer-go.org/mailman/listinfo/computer-go ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
Re: [Computer-go] A new ELF OpenGo bot and analysis of historical Go games
And most important: * Does ELF know the meaning of life? On 16/02/2019 17:29, "Ingo Althöfer" wrote: Hi Remi, thanks you for the link. A few questions (to all who know something): * How strong is the new ELF bot in comparison with Leela-Zero? * How were komi values taken into account when analysing old go games with help of ELF? * How often does ELF propose moves played by AlphaGo (for instance in the games with Fan Hui, Lee Sedol, and in the sixty games from December 2017)? * Does ELF understand that the strength of AlphaGo increased from October 2015 to May 2017? Cheers, Ingo. ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
Re: [Computer-go] A new ELF OpenGo bot and analysis of historical Go games
Hi Remi, thanks you for the link. A few questions (to all who know something): * How strong is the new ELF bot in comparison with Leela-Zero? * How were komi values taken into account when analysing old go games with help of ELF? * How often does ELF propose moves played by AlphaGo (for instance in the games with Fan Hui, Lee Sedol, and in the sixty games from December 2017)? * Does ELF understand that the strength of AlphaGo increased from October 2015 to May 2017? Cheers, Ingo. ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
[Computer-go] A new ELF OpenGo bot and analysis of historical Go games
Hi, I have just discovered that interesting Facebook blog post, in case anybody else missed it: https://ai.facebook.com/blog/open-sourcing-new-elf-opengo-bot-and-go-research/ Rémi ___ Computer-go mailing list Computer-go@computer-go.org http://computer-go.org/mailman/listinfo/computer-go
