yss at bd.mbn.or.jp
Sun May 10 04:39:32 PDT 2015
According to the author,
DCNNigo [5k] uses his own learning data.
Its input are 6 layers, Black, White, Lib1, Lib2, Lib3, Lib4.
There is no previous move info.
It takes 2.9ms on GTX670 cublas, accuracy is 37%.
DCNNigo2,3,4 [3k] uses Detlef Schmicker's binary. 44% accuracy.
He said, it takes 70ms per move.
Author is FUKUMOTO Kiyoshi-san, He is also an author of Hirabot.
----- Original Message -----
From: "Michael Sué" <michael.sue at gmx.de>
To: <computer-go at computer-go.org>
Sent: Sunday, May 10, 2015 8:14 PM
Subject: [Computer-go] DCNNigo*
> Lately I saw 4 bots DCNNigo* playing on KGS with different strength. What is the difference between them?
> - Michael.
> Computer-go mailing list
> Computer-go at computer-go.org
More information about the Computer-go