Jump to content

What happens when a geek feeds every current model into a recurrent neural net?

Recommended Posts

This, that's what.

I took the JSON file that @ZiggyQubert uses to build the cards in G.B.Scrum (check it out Android users, it's great) and fed them into a 3-layer recurrent neural network. 

After 6 hours or so of the RNN learning all about Guild Ball, I asked it to generate some players for a guild called "the dungeoneers".


The RNN spat out some very convincing models; some of the names are a bit crazy and I took some license in renaming a couple of plays that weren't pronounceable, but on the whole it did a good (if wacky) job.


The overall feel is of a team that does really weird things with the ball, takes lots of additional advances, and hits like a ton of bricks if you accidentally wander into the fairly low threat ranges of their beaters.  I especially like the spiteful mascot.




The rest of the guild is available as a PDF here

Share this post

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now