Mechanistic Tomography: Designed Measurement for Control-Oriented Interpretability

• URL: http://arxiv.org/abs/2608.19338v1
• Date: Wed, 19 Aug 2026 18:02:03 GMT
• Title: Mechanistic Tomography: Designed Measurement for Control-Oriented Interpretability
• Abstract summary: Mechanistic interpretability seeks quantities that models do not expose directly.<n>We formulate their shared measurement structure as mechanistic tomography.<n>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.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: 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.

Related papers

• Calibration is the Bottleneck: An Action-Class Diagnostic of Multi-Turn Tool-Calling [65.03665015319457]
  We propose an action-class-oriented diagnostic framework that decomposes multi-turn failures into two modes: action-class miscalibration and action-execution failure.<n>We validate it on a panel of tool-calling models across multiple multi-turn benchmarks.
  arXiv  Detail & Related papers  (2026-09-01T09:08:15Z)
• The Quadrilateral Loss: Additivity as a Measurable Behavior of Dense Neural Networks [0.0]
  We introduce the quadrilateral loss, a differentiable penalty that treats additivity as a measurable behavior instead.<n>We compare routes to exact additivity, spanning structural masks, behavioral penalties, weight decay, backfitting, the shared-section model, and bagged boosted stumps.
  arXiv  Detail & Related papers  (2026-07-22T14:23:37Z)
• DriftScope: Measuring The Hidden Effects of Diffusion Model Adaptation [50.584834483403675]
  Adapting pre-trained text-to-image diffusion models, whether to learn new visual concepts or erase unwanted ones, is routinely evaluated on its intended effects alone.<n>We demonstrate that adaptation systematically damages semantically unrelated concepts in ways that aggregate metrics structurally cannot surface.<n>We introduce DriftScope, a prompt-level diagnostic tool that takes any two model checkpoints and returns a ranked list of tokens whose visual concepts have shifted most between them.
  arXiv  Detail & Related papers  (2026-06-30T20:57:51Z)
• Calibration Is Not Control: Why LLM-Agent Oversight Needs Intervention [14.376426170653707]
  Two trajectory prefixes can have the same risk estimate while requiring different actions, because one remains recoverable and the other does not.<n>We formalize this mismatch as target error and identify intervention advantage, the expected utility gain from intervening rather than continuing.
  arXiv  Detail & Related papers  (2026-06-19T13:08:17Z)
• Input-schema identifiability limits in physics-informed surrogates for mechanics-governed flow [0.0]
  We introduce an input-schema identifiability certificate for computational surrogates.<n>We instantiate the framework for incompressible tubular flow using a Cosserat-rod reduction.<n>The results provide a mechanics-based diagnostic for deciding whether a surrogate modelling task is physically identifiable.
  arXiv  Detail & Related papers  (2026-06-08T23:22:18Z)
• Measurement Geometry and Design for Trustworthy Generative Inverse Problems [9.927046475467204]
  We study generative inverse problems from a measurement-geometry perspective.<n>The central question is whether a fixed measurement operator can distinguish nearby images that are plausible under the generative prior.<n>We introduce a local measurement-manifold compatibility measure that quantifies how well the operator observes prior-relevant tangent directions.
  arXiv  Detail & Related papers  (2026-06-01T14:28:24Z)
• An Exponential Advantage for Adaptive Tomography of Structured States under Pauli Basis Measurements [20.661777397879792]
  Broad claims about whether adaptivity helps in quantum state tomography can be misleading unless the state family, measurement architecture, and error metric are specified carefully.<n>We study a restricted but physically important regime: single-copy quantum state tomography under local Pauli basis measurements.
  arXiv  Detail & Related papers  (2026-04-28T18:26:47Z)
• On the Paradoxical Interference between Instruction-Following and Task Solving [50.75960598434753]
  Instruction following aims to align Large Language Models (LLMs) with human intent by specifying explicit constraints on how tasks should be performed.<n>We reveal a counterintuitive phenomenon: instruction following can paradoxically interfere with LLMs' task-solving capability.<n>We propose a metric, SUSTAINSCORE, to quantify the interference of instruction following with task solving.
  arXiv  Detail & Related papers  (2026-01-29T17:48:56Z)
• Nonparametric Identifiability of Causal Representations from Unknown Interventions [63.1354734978244]
  We study causal representation learning, the task of inferring latent causal variables and their causal relations from mixtures of the variables. Our goal is to identify both the ground truth latents and their causal graph up to a set of ambiguities which we show to be irresolvable from interventional data.
  arXiv  Detail & Related papers  (2023-06-01T10:51:58Z)
• Score-based Causal Representation Learning with Interventions [54.735484409244386]
  This paper studies the causal representation learning problem when latent causal variables are observed indirectly. The objectives are: (i) recovering the unknown linear transformation (up to scaling) and (ii) determining the directed acyclic graph (DAG) underlying the latent variables.
  arXiv  Detail & Related papers  (2023-01-19T18:39:48Z)
• Uncertainty-Aware Adaptation for Self-Supervised 3D Human Pose Estimation [70.32536356351706]
  We introduce MRP-Net that constitutes a common deep network backbone with two output heads subscribing to two diverse configurations. We derive suitable measures to quantify prediction uncertainty at both pose and joint level. We present a comprehensive evaluation of the proposed approach and demonstrate state-of-the-art performance on benchmark datasets.
  arXiv  Detail & Related papers  (2022-03-29T07:14:58Z)
• Robust Self-Supervised LiDAR Odometry via Representative Structure Discovery and 3D Inherent Error Modeling [67.75095378830694]
  We develop a two-stage odometry estimation network, where we obtain the ego-motion by estimating a set of sub-region transformations. In this paper, we aim to alleviate the influence of unreliable structures in training, inference and mapping phases. Our two-frame odometry outperforms the previous state of the arts by 16%/12% in terms of translational/rotational errors.
  arXiv  Detail & Related papers  (2022-02-27T12:52:27Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.