AI News HubLIVE
Original source2 min read

Targeted Recovery of Weight-Space Mechanisms From Neural Networks

Parameter decomposition (PD) decomposes neural networks into interpretable components but is computationally expensive for large models. The proposed targeted PD (tPD) introduces a high-rank catch-all component to handle non-target data, enabling efficient recovery of circuits for specific inputs. tPD extracts a CSS-only submodel from a 4-block transformer using 7% of the FLOPs of published decomposition, and surgically ablates memorized sequences in a 12-block transformer with negligible side effects. Accepted at the Mechanistic Interpretability Workshop, ICML 2026.

SourcearXiv Machine LearningAuthor: Antoine Vigouroux, Lee Sharkey

-->

[Submitted on 19 Jun 2026]

Title:Targeted Recovery of Weight-Space Mechanisms From Neural Networks

View a PDF of the paper titled Targeted Recovery of Weight-Space Mechanisms From Neural Networks, by Antoine Vigouroux and 1 other authors

View PDF HTML (experimental)

Abstract:Parameter decomposition (PD) decomposes neural networks into interpretable computational components that faithfully reflect the original network's operations. However, scaling PD to large models requires vast compute, making it a costly and risky endeavor. Here we propose targeted PD (tPD), which identifies only the components that process specific inputs of interest -- from isolated prompts to large subtasks -- by introducing a high-rank catch-all component that handles all non-target data. We validate tPD on toy models and on transformer language models trained on The Pile, where it recovers reproducible, mechanistically faithful circuits. We extract a CSS-only submodel of a 4-block transformer using 7% of the FLOPs of its published decomposition, and in a 12-block transformer we surgically ablate and rewire memorized sequences, with negligible side effects on other inputs.

Comments: Accepted at the Mechanistic Interpretability Workshop, ICML 2026

Subjects:

Machine Learning (cs.LG)

Cite as: arXiv:2607.13047 [cs.LG]

(or arXiv:2607.13047v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2607.13047

arXiv-issued DOI via DataCite

Submission history

From: Antoine Vigouroux [view email] [v1] Fri, 19 Jun 2026 19:07:38 UTC (3,565 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Targeted Recovery of Weight-Space Mechanisms From Neural Networks, by Antoine Vigouroux and 1 other authors

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.LG

new | recent | 2026-07

Change to browse by:

cs

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

IArxiv recommender toggle

IArxiv Recommender (What is IArxiv?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)