Interpreting Language Model Parameters

A new arXiv paper (2610.20748v1) introduces adVersarial Parameter Decomposition (VPD), a method that decomposes a language model's parameters into simple subcomponents each implementing only a small part of the model's learned algorithm, while requiring only a small fraction of those subcomponents to account for the network's behavior on any input. VPD optimizes decompositions that preserve the network's input-output behavior even when many subcomponents are ablated, including adversarially selected ablations, enabling a new kind of circuit analysis of how parameter subcomponents interact to produce outputs. The authors state more work remains to deepen understanding of how neural networks use parameters to compute behavior, but the approach suggests identifying a small set of mechanistically faithful subcomponents for further analysis.

arXiv:2610.20748v1 Announce Type: new Abstract: Mechanistic interpretability aims to uncover how neural networks use their parameters to implement their impressive neural algorithms. Although previous work has uncovered substantial structure in the intermediate representations that networks use, little progress has been made to understand how the parameters and nonlinearities of networks perform computations on those representations. In this work, we present a method that brings us closer to this understanding by decomposing a language model's parameters into subcomponents that each implement only a small part of the model's learned algorithm, while simultaneously requiring only a small fraction of those subcomponents to account for the network's behavior on any input. The method, adVersarial Parameter Decomposition VPD , optimizes for decompositions of neural network parameters into simple subcomponents that preserve the network's input-output behavior even when many subcomponents are ablated, including under ablations that are adversarially selected to destroy behavior. This encourages learning subcomponents that provide short, mechanistically faithful descriptions of the network's behavior that should aggregate appropriately into more global descriptions of the network's learned algorithm. We study how sequences of interactions between these parameter subcomponents produce the network's output on particular inputs, enabling a new kind of 'circuit' analysis. While more work remains to be done to deepen our understanding of how neural networks use their parameters to compute their behavior, our work suggests an approach to identify a small set of simple, mechanistically faithful subcomponents on which further mechanistic analysis can be based.