Generable: startup productizing generative models


#1

Hey all,

I just wanted to announce the startup I’m working on with @ericnovik: Generable. We’re productizing generative models, written in Stan, for industrial use.

Here’s a little bit about why we’ve built the company: http://www.generable.com/2017/10/27/why-generable.html

Please reach out if you’re interested. Thanks!


#2

Good to know and good luck. I like the “It’s about big models, not big data”, since some field of science have still not all their feet in the Big Data Era :-)


#3

I think it’s the right focus for Stan.

It’s been an easy sell to those wanting to fit complex models, but a hard sell to funding agencies like DoD and NSF.

Once we get MPI parallelism in place and basic GPU operations, we’ll be much better poised to handle much bigger data. But we’ll still want to do full Bayes. Maybe ADVI will work much better in a classic big-data setting.