Computational Mechanics Hackathon (June 1 & 2)

post by Adam Shai (adam-shai) · 2024-05-24T22:18:44.352Z · LW · GW · 5 comments

Contents

5 comments

Join our Computational Mechanics Hackathon, organized with the support of APART, PIBBSS and Simplex


This is an opportunity to learn more about Computational Mechanics, its applications to AI interpretability & safety, and to get your hands dirty by working on a concrete project together with a team and supported by Adam & Paul. Also, there will be cash prizes for the best projects!
 

Read more and sign up for the event here

We’re excited about Computational Mechanics as a framework because it provides a rigorous notion of structure that can be applied to both data and model internals. In, Transformers Represent Belief State Geometry in their Residual Stream [LW · GW] , we validated that Computational Mechanics can help us understand fundamentally what computational structures transformers implement when trained on next-token prediction - a belief updating process over the hidden structure of the data generating process. We then found the fractal geometry underlying this process in the residual stream of transformers.
 

This opens up a large number of potential projects in interpretability. There’s a lot of work to do!

Key things to know: 

5 comments

Comments sorted by top scores.

comment by Alex_Altair · 2024-05-25T01:58:25.044Z · LW(p) · GW(p)

I'm curious if you knowingly scheduled this during LessOnline?

Replies from: adam-shai, adam-shai, Sodium
comment by Adam Shai (adam-shai) · 2024-05-29T15:55:24.740Z · LW(p) · GW(p)

We've decided to keep the hackathon as scheduled. Hopefully there will be other opportunities in the future for those that can't make it this time!

comment by Adam Shai (adam-shai) · 2024-05-25T14:17:33.318Z · LW(p) · GW(p)

No, thanks for pointing this out

comment by Sodium · 2024-05-25T13:13:35.662Z · LW(p) · GW(p)

It's also EAG London weekend lol it's a busy weekend for all

Replies from: adam-shai
comment by Adam Shai (adam-shai) · 2024-05-25T14:21:25.487Z · LW(p) · GW(p)

Also a good point. Thanks