Chessbase claims that their Fat Fritz differs from other NN in that it also has an evalutation function.
How does eval come into play, given that NN is based on self play and weighted averages?
How does eval come into play, given that NN is based on self play and weighted averages?
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill