pith. sign in

arxiv: 2412.03773 · v1 · pith:FS63F4S3 · submitted 2024-12-04 · cs.LG · cs.AI

Modular addition without black-boxes: Compressing explanations of MLPs that compute numerical integration

pith:FS63F4S3open to challenge →

classification cs.LG cs.AI
keywords compressingmodelsadditionfeature-mapsmodularnonlinearreluwork
0
0 comments X
read the original abstract

The goal of mechanistic interpretability is discovering simpler, low-rank algorithms implemented by models. While we can compress activations into features, compressing nonlinear feature-maps -- like MLP layers -- is an open problem. In this work, we present the first case study in rigorously compressing nonlinear feature-maps, which are the leading asymptotic bottleneck to compressing small transformer models. We work in the classic setting of the modular addition models, and target a non-vacuous bound on the behaviour of the ReLU MLP in time linear in the parameter-count of the circuit. To study the ReLU MLP analytically, we use the infinite-width lens, which turns post-activation matrix multiplications into approximate integrals. We discover a novel interpretation of} the MLP layer in one-layer transformers implementing the ``pizza'' algorithm: the MLP can be understood as evaluating a quadrature scheme, where each neuron computes the area of a rectangle under the curve of a trigonometric integral identity. Our code is available at https://tinyurl.com/mod-add-integration.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Interactions Between Crosscoder Features: A Compact Proofs Perspective

    cs.LG 2026-06 unverdicted novelty 6.0

    Derives an interaction measure between crosscoder features from reconstruction error in compact proofs and applies it to produce computationally sparse crosscoders retaining 60% MLP performance with single-feature sel...