- 31 Jul, 2020 2 commits
- 30 Jul, 2019 4 commits
- 16 Sep, 2016 2 commits
- 20 Jun, 2014 2 commits
- 07 Dec, 2013 3 commits
- 24 Nov, 2013 6 commits
- 22 Nov, 2013 6 commits
-
-
chaz authored
It's a more accessible way to describe what's going on.
-
chaz authored
-
chaz authored
Seems to work pretty good an' stuff.
-
chaz authored
-
chaz authored
-
chaz authored
This changes the way the next generation is populated. Instead of cloning and mutating the top performers in the last generation, we select two individuals, preferring the top performers, and combine their genome (and mutate the result). This results in much wider coverage in the search for better individuals.
-
- 21 Nov, 2013 5 commits
- 20 Nov, 2013 1 commit
-
-
chaz authored
-
- 17 Nov, 2013 7 commits
-
-
chaz authored
-
chaz authored
-
chaz authored
-
chaz authored
-
chaz authored
The graph was pretty pointless with how scoring ended up looking. Now we just print some info about the last run so you know it's still working, even when the top score stays the same for generations.
-
chaz authored
Instead of having each output node be one of two values, this makes each output node have the sum of its input. If you want the old behavior, just add an extra layer with the proper thresholds and weights.
-
chaz authored
-
- 08 Oct, 2013 2 commits