Dmitry Vaintrob

We're posting lots of papers written with GPT-6 Astra on generalization properties of bayesian neural net learning. [1] The following picture emerges: * For various reasonable priors on MLP weights, [2] the induced prior on functions is a circuit prior: for fixed depth, the ordering of functions by negative log...

I'm sharing preliminary results of a suite of experiments I ran with claudecode on a small LLM (gpt2-small, no Layer Norm version, courtesy of Apollo research. most of these are on the layer-6 MLP). The github repo for the experiments is here. The success of these experiments given the method's...

Here are the slides of a talk Kaarel gave, presenting work with Dmitry establishing that (even arbitrarily overparametrized) neural net bayesian learning has a circuit prior — and thus, when learning a function which is implemented by some small circuit, only requires a small amount of training data to get...

This is a post explaining my paper with Kaarel Hänni on complexity of infinite-width networks. I will explain the result, why it matters, and how the mathematical idealizations can interact with real structure in neural nets. This leads to some threads I am excited to pull on more in the...

In a recent post, we presented PIRAMID, its leadership and research pillars, and a plan for how they fit together. In this post, we sketch a team-by-team account of progress and targets for the next 6–12 months. We include results to date as evidence of viability: we’ve been a small...

Principles of Intelligence (PrincInt, formerly PIBBSS) is launching PIRAMID, an internal research division using the tools and techniques of statistical physics to build scientific foundations for ambitious mechanistic interpretability. PIRAMID’s central premise is that scalable alignment will require more than persuasive ad-hoc explanations of model behavior. It will require interpretability...

A correction: Sumio Watanabe pointed out in an email that the RLCT of the learning zero function scales like rather than just , as shown in this paper by Aoyagi and Watanabe (unfortunately not openly available). I generalize this result to more general networks here, as well as give a...