Deep Pepper Paper

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

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
Leo
Posts: 629
Joined: Fri Sep 16, 2016 4:55 pm
Location: Saint Paul, Minnesota USA

Deep Pepper Paper

Post by Leo » Sat Jul 07, 2018 4:53 pm

Does anyone want to comment on this paper?

https://export.arxiv.org/pdf/1806.00683
AMD fan.

jdart
Posts: 3663
Joined: Fri Mar 10, 2006 4:23 am
Location: http://www.arasanchess.org

Re: Deep Pepper Paper

Post by jdart » Sat Jul 07, 2018 5:44 pm

I am not sure I understand all of it but in general it looks like a plausible approach.

They mention they don't have an ELO measure based on performance against other engines. So that makes it hard to judge how effective a system it is.

I also don't know if they have managed to overcome to tactical blindness that LC0 exhibits on occasion.

--Jon

whereagles
Posts: 547
Joined: Thu Nov 13, 2014 11:03 am

Re: Deep Pepper Paper

Post by whereagles » Sun Jul 08, 2018 3:13 pm

no games, no fun, no matter :roll:

saikrishnagv
Posts: 2
Joined: Mon Sep 24, 2018 7:42 pm
Full name: Sai Krishna

Re: Deep Pepper Paper

Post by saikrishnagv » Tue Sep 25, 2018 5:26 pm

Hello gang,

I am one of the authors of the paper, let me know if you have any questions. Our paper is an attempt for much faster training of self-play like algorithms. Self-play algorithms were proposed by David Barber's group https://arxiv.org/abs/1705.08439 and then followed by some papers from DeepMind

-Sai

Post Reply