Metraipolis › archive › dc_aa044c8b769a

Mechanistic Tomography: Designed Measurement for Control-Oriented Interpretability — arXiv:2608.19338

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.19338
License (as archived): CC0-1.0
Archived: 2026-10-18T11:48:06.833Z
SHA-256 of the archived text: b18e1845b248875d3daff6c015baf25720b3a688bd9a3087cf32aed8b4dd918f
Stable URL of this page: https://metraipolis.com/documento/dc_aa044c8b769a

How to cite

Mechanistic Tomography: Designed Measurement for Control-Oriented Interpretability — arXiv:2608.19338 — arXiv (en), https://arxiv.org/abs/2608.19338 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_aa044c8b769a (Metraipolis archive dc_aa044c8b769a, sha256 b18e1845b248875d3daff6c015baf25720b3a688bd9a3087cf32aed8b4dd918f)

BibTeX

@misc{dcaa044c8b769a,
  title = {Mechanistic Tomography: Designed Measurement for Control-Oriented Interpretability — arXiv:2608.19338},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.19338},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_aa044c8b769a; sha256 b18e1845b248875d3daff6c015baf25720b3a688bd9a3087cf32aed8b4dd918f},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.19338. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Mechanistic interpretability seeks quantities that models do not expose directly: represented states, component effects, interactions, and responses to interventions. Patching, gradients, Hessian-vector products, and subset interventions provide different measurements under different access assumptions and may target different quantities. We formulate their shared measurement structure as mechanistic tomography: designed measurement for recovering internal mechanisms and intervention effects. For a chosen basis and intervention family, measurements take the form y = Ax + w, where A describes the interventions, x is the target map, and w contains nonlinear response, sampling error, and basis misspecification. This language gives a practical procedure: start with the least costly measurements, test on held-out interventions at the intended scale, calibrate simple mismatch, and expand the measurement family when structured residuals remain. Control provides a demanding validation setting because an estimate that guides an intervention acts as an observer. In a two-HMM model, control error rises with observer error, while target improvement can hide nuisance-state movement. Under forward-only access, sparse aggregate measurements recover a finite-effect map with fewer interventions than coordinate patching. With gradient access, finite probes improve a local attribution map. Lifted measurements and Hessian-vector products recover interactions missed by first-order maps, while Tracr shows that the required family depends on the basis. On GPT-2-small IOI, the Name Mover-Negative Name Mover interaction is the largest held-out predictive term among three tested cross-group pairs. On Qwen-2.5-7B, finite calibration makes an additive refusal-response map adequate, so held-out error does not support pairwise lifting.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-21 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.19338 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-18T11:47:43.293Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-21?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).