Mechanistic Interpretability

Follow this topic Ask about this topic

The field seeks to explain how a model’s internal computations produce particular outputs, rather than relying only on input-output behavior. Researchers use sparse autoencoders and transcoders to identify features, and probes, attribution graphs, activation patching, ablation, and steering to trace or test their roles. Studies examine language-model tasks such as factual recall, reasoning, and refusal, as well as vision, audio, and scientific models. A central challenge is that a signal can be easy to decode without being causally used, while circuit explanations and interventions can vary across prompts, models, and analysis choices.

Latest papers

391 papers63 papers in the last four weeks, up 75% on the four weeks before. 0.4% of all new papers.

May 23, 2026

Measuring Alignment-Induced Activation Shifts Correctly: A Template-Controlled Difference-in-Differences Protocol

Comparing a model's internal activations before and after alignment is a natural way to ask what safety training changes: one forms the matrix of paired aligned-minus-base activations on safety-relevant inputs and reads off its effective rank or top direction. We show the obvious way to form this matrix is confounded. The aligned model is evaluated under a chat template the base model never saw, so the naive difference conflates the alignment shift with chat formatting. We introduce a four-variant decomposition of the modification matrix (naive, template-controlled, within-aligned, and difference-in-differences, DiD) that separates the two effects. Template control alone removes a 2.0-3.9x inflation of the measured effective rank across Llama-3.1-8B, Gemma-2-9B, and Qwen-2.5-7B; the DiD contrast is what recovers the refusal direction of Arditi et al. (2024), lifting its cosine alignment from 0.18-0.39 to 0.50-0.86. Projection-ablation across the three families confirms the recovered subspace is behaviorally active and that singular-value order is not causal order. We validate the protocol on a controlled testbed and distill it into measurement recommendations for activation-difference studies of alignment.

Yuki Nakamura

The Open University of Japan

LLM Alignment Mechanistic Interpretability Causal Inference

May 23, 2026

Polymorphism Is Rotation: Operational Mechanistic Interpretability from a Two-Layer Transformer to Pythia-70m

Independently trained transformers compute the same function in residual-stream bases that differ by a uniform random rotation on

\mathrm{SO}(d_{\mathrm{model}})

. We call this phenomenon polymorphism: same function, mutually unintelligible interior coordinates. One matrix multiplication per model pair removes it: an orthogonal Procrustes fit on a single batch of activations transfers sparse-autoencoder feature dictionaries and steering vectors between independently trained models, with no retraining. The phenomenon is invisible to the standard SAE universality metric. Decoder-column cosine similarity matches across seeds at 98%, the SAE-universality headline number, while an SAE trained on one seed reconstructs another seed's activations at negative explained variance, worse than predicting the constant mean. The decoder columns align; the encoder reads from a rotated frame. A single Procrustes rotation

R

restores reconstruction to within 0.025 EV of the within-seed ceiling at every internal site.

R

is Haar-distributed:

\|R - I\|_F

matches the random-orthogonal prediction

\sqrt{2 d_{\mathrm{model}}}

to 0.1% at

d_{\mathrm{model}} = 512

, and a Kolmogorov-Smirnov test of

R

's eigenvalue spectrum against Haar

\mathrm{SO}(d_{\mathrm{model}})

returns

p \approx 1.000

pooled and per-pair. Diff-of-means steering vectors transfer in three regimes by alignment with

R

's invariant subspace: clean when pinned by shared output weights, partial when overlapping the rotated subspace, inverted otherwise. With no shared I/O (Pythia), all three collapse to universally inverted. The same rotation account holds across training checkpoints within a single run. Validated on a 104k-parameter Dyck-3 transformer and nine independently-trained Pythia-70m seeds on The Pile, via a pre-registered four-bar operational framework. Frontier-scale (10B+) replication remains open.

Jordan F. McCann

Independent Researcher

Orthogonal Procrustes Alignment Transformer Interpretability Transformer Sparse Autoencoders Mechanistic Interpretability Representation Alignment+1

May 22, 2026

Every Component Is a Lookup: One Linear Graph for Interaction, Composition and Attribution

Interpretability methods for transformers are typically built around separate questions: which components interact, how information routes to the output, and which input tokens contribute. Because these methods rely on different assumptions, their answers are difficult to relate. We argue that two architecturally motivated assumptions suffice to address all three questions: attention and MLPs share a key-value form,

φ(S)\,U

, in which

φ(S)

selects over values

U

, and components read from an additive residual stream, the sum of component outputs. Holding these selections at their forward-pass values turns the model into a computational graph, of which component interactions, composition paths, and token attribution are different readouts. We develop Unpack, a backward attribution procedure over this graph, and validate each readout against the corresponding established test: interaction scores predict ablation effects across models from 160M to 6.9B parameters, recovered routes reproduce established circuits down to the key, query, or value branch the circuit specifies, and token attribution passes the same faithfulness test as dedicated attribution methods. The results suggest that these two assumptions suffice for the interpretability questions above. On a task with a known circuit, we find that contribution and causal effect can differ, and that the difference has a recognisable signature: components that matter for the task change their contribution when the task is removed from the input, while components that act like a bias term do not. Code is available at https://github.com/Fun-Cry/unpacklm.

Po-Kai Chen, Aske Plaat, Niki van Stein

Leiden University

Transformer Interpretability Token-Level Credit Assignment Mechanistic Interpretability Fine-Grained Credit Assignment

May 22, 2026

Spectral Probe-Circuits: A Three-Step Recipe for Identifying Attention-Head Circuits in Pretrained Transformers

We present a three-step recipe for identifying attention-head circuits in pretrained transformers. A per-head spectral signal -- the time-integrated participation ratio of each head's attention output -- ranks heads doing sustained content-dependent computation without labels or attribution gradients. A task-pattern screen filters this general indicator into a task-specific candidate circuit, and group ablation against a matched-random control completes the causal claim. We validate across an 8x parameter range (51M to 1B-active / 7B-total), two architecture families (dense, mixture-of-experts), and four pretraining pipelines. The recipe ports: a 2-6 head induction circuit is causally necessary in every model tested, with a 94-100% drop in synthetic-induction top-1 after ablation. The spectral signal is predictive without supervision: on six independent seeds of a 51M-parameter probe model, the same computation identifies the seed-specific circuit on each seed. The fraction of heads doing identifiable specialized computation is conserved at 17-19% across the Pythia family (124M to 410M), while specific induction circuits stay 3-11 heads -- sublinear in total head count. This paper is the methodology anchor of a three-paper program; companion papers extend the recipe to developmental trajectories during pretraining and to composed-task circuits where pattern selectivity decouples from task-causal structure.

Yongzhong Xu

Transformer Interpretability Circuit Discovery Attention Head Analysis Self-Attention Mechanistic Interpretability Induction Heads+2

May 21, 2026

Mechanistic origins of catastrophic forgetting: why RL preserves circuits better than SFT?

Fine-tuning large language models (LLMs) frequently induces catastrophic forgetting of prior capabilities. Recent work has shown that reinforcement learning (RL) retains prior capabilities more effectively than supervised fine-tuning (SFT), attributing this to policy-gradient updates remaining closer to the base policy \cite{shenfeld2025rl}. We extend this behavioral account to the mechanistic level and ask whether RL's advantage is mirrored by stronger preservation of internal computational circuits. We introduce differential circuit vulnerability, a head-level measure of how much a circuit degrades under fine-tuning, and use it to compare RL and SFT on Qwen2.5-3B-Instruct adapted to scientific question-answering. We find a clear mechanistic trade-off: SFT adapts more rapidly to the target task but produces substantially greater circuit disruption and forgetting of prior capabilities, whereas RL preserves a larger fraction of the base circuit at the cost of slower task adaptation. These findings suggest that circuit preservation may help explain why RL is more robust to catastrophic forgetting. We released our code here: https://github.com/rl-sft-circuit-research/differential-circuit-vulnerability.

Jeanmely Rojas Nunez, Viraj Sawant, Nathan Allen +4

Algoverse AI Research · Independent

Supervised Fine-Tuning Mechanistic Interpretability Circuits in Language Models RL Fine-Tuning Catastrophic Forgetting

May 21, 2026

Transcoders Trace Visual Grounding and Hallucinations in Vision-Language Models

Generative Vision-Language Models (VLMs) perform well on multimodal reasoning, but how visual inputs are transformed to text remains poorly understood. Existing interpretability work on VLMs uses Sparse Autoencoders (SAEs), which decompose static residual representations and miss the functional updates that drive cross-modal interaction. We adopt a function-centric framework based on Transcoders, sparse approximations of MLP sublayers that act as a causal proxy for layer-wise computation. Applied to Gemma 3-4B-IT, the framework decomposes the model into interpretable computational pathways linking image patches to directions in token generation. Transcoder attributions produce stronger and more stable effects on visually grounded tokens under patch ablation than SAE attributions, and align better with semantically relevant image regions. A False Visual Grounding counterfactual analysis confirms that the recovered pathways are specific to vision-language interaction.Finally, we perform a structural analysis of hallucinated generations, by extracting graph-based indicators from circuit traces produced by the transcoders. A logistic classifier over these mechanistic graph features predicts hallucinations at AUC

0.68

. These results show that function-centric circuit decomposition yields interpretable and predictive accounts of multimodal computation in VLMs.

Dimitrios Damianos, Leon Voukoutis, Georgios Skyrianos +2

Institute of Language and Speech Processing, Athena Research Center Athens, Greece

Object Hallucination in VLMs VLM Interpretability Mechanistic Interpretability Vision-Language Grounding Multimodal Model Interpretability

May 21, 2026

Towards Verifiable Transformers: Solver-Checkable Circuit Explanations

Mechanistic interpretability often identifies circuits inside Transformer models, but explanations of those circuits are usually validated through examples, ablations, and manual reasoning. This leaves a gap between finding a plausible circuit and proving what the circuit does. We introduce Verifiable Transformers, a framework for converting task-localized Transformer circuits into bounded, solver-checkable claims. Given a behavior, a finite task domain, and a candidate-token projection, we extract a task circuit and verify properties such as projected functional equivalence, edge necessity, task-relevant invariance, and final-residual robustness. Direct verification encodes the extracted circuit itself into an SMT solver. When a circuit contains operators that are not exactly or tractably encodable, surrogate-mediated verification fits an SMT-encodable surrogate, validates it against the extracted circuit over the bounded domain, and verifies symbolic explanations against the surrogate. We instantiate direct verification with a GPT-style architecture using Signed L1 BandNorm, sparsemax attention, and LeakyReLU. On small symbolic sequence tasks, we train an SMT-representable Transformer, extract sparse circuits for quote closing and bracket type tracking, and exhaustively verify projected functional equivalence, content invariance, edge necessity, and final-residual robustness. At GPT-2 scale, the same operator stack trains stably on OpenWebText, although naive direct SMT verification remains intractable. We also demonstrate surrogate-mediated verification on task-localized circuits with hard-to-encode attention, showing both verified symbolic explanations and solver-generated counterexamples. The goal is not full-model verification, but a concrete path for turning mechanistic circuit explanations into formal propositions that can be proven or refuted.

Neel Somani

Independent Researcher

Neural Network Verification Transformer Mechanistic Interpretability

May 21, 2026

Geometry-Adaptive Explainer for Faithful Dictionary-Based Interpretability under Distribution Shift

Mechanistic interpretability aims to explain a model's behavior by identifying causally responsible internal structures. Dictionary-based explainers such as sparse autoencoders and transcoders are a primary tool, but their faithfulness under out-of-distribution (OOD) shift has received little systematic attention. We show that distribution shift rotates the subspace that the model actively uses, misaligning the explainer's dictionary trained on in-distribution (ID) activations. We formalize this misalignment as the faithfulness gap, a geometric distance between the ID dictionary and the OOD-active subspace, and show that it controls OOD faithfulness degradation. To reduce this gap, we propose the Geometry-Adaptive Explainer (GAE), which realigns the explainer's dictionary with the OOD-active subspace while preserving the original feature structure. This requires only unlabeled OOD activations and no gradient updates. We prove that GAE improves over the unadapted ID explainer, with excess loss bounded quadratically by the second-moment shift. Empirically, GAE even matches or surpasses all training-based baselines in causal faithfulness across multiple models and OOD settings.

Sungjun Lim, Heedong Kim, Andrew Lee +1

aYonsei University · bHarvard University

Distribution Shift Robustness OOD Generalization Mechanistic Interpretability Causal Interpretability Explainability Evaluation

May 20, 2026

From Circuit Evidence to Mechanistic Theory: An Inductive Logic Approach

Mechanistic interpretability produces circuit-level causal analyses of neural network behaviour, but discovered circuits often remain isolated experimental artefacts: there is no shared formal representation for what circuits compute, how they relate, or when two findings provide evidence for the same mechanism. This work provides a formal infrastructure for cumulative mechanistic science by treating circuit interpretation as inductive theory construction. Each circuit is characterised at two levels: a Causal Functional Signature (CFS), which grounds component behaviour in causal attribution evidence and token role profiles, and an architectural signature

τ_{\mathrm{arch}}

, learned by inductive logic programming (ILP) from scale-invariant structural predicates. Together, these constitute a formal coherence layer that makes mechanistic claims explicit, comparable via

θ

-subsumption, and portable across model scales. CFS reveals qualitatively distinct computational strategies across task types, including attention-mediated copying versus MLP-mediated binding. ILP signatures achieve substantially better structural separation than graph kernel and feature-vector baselines, and support principled transfer across model scales and architecture families.

Nura Aljaafari, Danilo S. Carvalho, Andre Freitas

Department of Computer Science, University of Manchester, United Kingdom · CRUK National Biomarker Centre, University of Manchester, United Kingdom · Idiap Research Institute, Switzerland

Circuit Discovery Mechanistic Interpretability Inductive Logic Programming Causal Interpretability

May 20, 2026

A Mechanistic Study of Tabular Foundation Models

Tabular foundation models with different architectures converge in accuracy across a range of classification and regression tasks. This raises questions a leaderboard cannot answer: (i) whether the models execute the same in-context algorithm, (ii) where row, column, and class-permutation invariances originate, and (iii) how robust they are under perturbations engineered against the inferred mechanism. We characterize all three. The model families realize qualitatively distinct similarity-based readouts: from an attention-weighted vote over context labels to a class-conditional mean readout, each confirmed by causal intervention. We find that the representation collapse highlighted in prior work is not a practical concern for them. Each model's permutation invariances trace to specific positional parameters whose removal preserves accuracy and makes approximate invariance exact. Perturbations engineered against each readout reproduce predicted failure modes; hub and rank attacks isolate them from refit baselines. Together these results give a mechanistic account of contemporary tabular foundation models and identify which inductive biases govern both their accuracy and characteristic failures.

Marin Biloš, James T. Wilson, Anderson Schneider +1

Machine Learning Research, Morgan Stanley

Tabular Foundation Models Mechanistic Interpretability Invariant Representation Learning Tabular ICL

May 20, 2026

Mechanistic Interpretability for Learning Assurance of a Vision-Based Landing System

EASA's learning-assurance guidance requires data-driven aviation systems to build and monitor their own situation representation, yet for neural networks the technical means to provide such evidence remain an open problem. We address this gap for a vision-based aircraft landing system: we propose that a minimally assurable model must at least be shown to separate content from style in its own situation representation. Showing that the model's predictions then rely largely on the contentful representation components leads to a concrete assurance path. To demonstrate this assurance path on a concrete model we train a vision transformer model for runway keypoint regression on the LARDv2 dataset. The model, which acts as the subject for our assurance demonstration, produces per-patch embeddings that we decompose into interpretable atoms via K-SVD sparse dictionary learning. A qualitative visualization confirms that contentful atoms track task-relevant runway structure and stylistic atoms track domain-specific appearance, and the regression head is shown to place almost all of its linear weight on contentful atoms. We further build on the content/style separation and define out-of-model-scope (OOMS) detection, a novel runtime assurance approach directly monitoring the model's situation representation. OOMS monitoring is complementary to operational design domain and output-space out-of-distribution monitoring and addresses concrete requirements of the recent EASA guidance. By directly analyzing a model's situation representation both at test time and runtime, this work delivers the first concrete piece of the representation-level evidence that EASA learning-assurance guidance demands, and points to mechanistic interpretability as a practical building block of future aviation safety cases.

Romeo Valentin, Olivia Beyer Bruvik, Marc R. Schlichting +1

Stanford Intelligent Systems Laboratory, Stanford University, Stanford, CA, USA

Aviation Safety Representation Disentanglement Mechanistic Interpretability Runtime Assurance

May 18, 2026

Language-Switching Triggers Take a Latent Detour Through Language Models

Backdoor attacks on language models pose a growing security concern, yet the internal mechanisms by which a trigger sequence hijacks model computations remain poorly understood. We identify a circuit underlying a language-switching backdoor in an 8B-parameter autoregressive language model, where a three-word Latin trigger (nine tokens) redirects English output to French. We decompose the circuit into three phases: (1) distributed attention heads at early layers compose the trigger tokens into the last sequence position; (2) the resulting signal propagates through mid-layers in a subspace orthogonal to the model's natural language-identity direction; (3) the MLP at the final layer converts this latent signal into French logits. The entire circuit flows through a serial bottleneck at a single position: corrupting that position at any layer entirely mitigates the trigger but also hinders the model's capabilities. The orthogonal latent encoding suggests that defenses that search for language-like signals in intermediate representations would miss this trigger entirely.

Francis Kulumba, Wissam Antoun, Théo Lasnier +2

1Inria Paris · 2Sorbonne Université

LLM Backdoor Attacks Mechanistic Interpretability Circuits in Language Models

May 18, 2026

Aligned Training: A Parameter-Free Method to Improve Feature Quality and Stability of Sparse Autoencoders (SAE)

Sparse autoencoders (SAEs) are one of the main methods to interpret the inner workings of deep neural networks (DNNs), decomposing activations into higher-dimensional features. However, they exhibit critical shortcomings where a large fraction of features are never activated and are unstable. Despite variants of SAEs that attempt to mitigate these issues, they require additional data, resampling, or training. We propose the aligned training, a parameter-free reparameterization of SAEs that simultaneously improves reconstruction quality, eliminates dead features, and significantly enhances stability across training seeds. Our approach is motivated by an overlooked observation that SAE feature quality, measured by the inner product between encoder and decoder directions (which we call the alignment score), follows a bimodal distribution across all modern architectures. The proposed aligned training enforces a geometric constraint between the encoder and decoder such that their inner product equals one for every feature, which removes a source of degeneracy in the SAE training without adding any hyperparameters. Across multiple models, dictionary sizes, and sparsity levels, the aligned training shows Pareto improvements on the SAEBench benchmarks. Beyond improving dead features, stability and reconstruction, our method readily integrates with techniques in mechanical interpretability such as Top/BatchTop-K architectures and p-Annealing. Overall, the aligned training substantially improves feature quality and stability of SAE without computational complexity or cost.

Michał Brzozowski, Neo Christopher Chung

Samsung AI Center, Warsaw, Poland · University of Warsaw, Poland

Sparse Autoencoders Mechanistic Interpretability Representation Alignment

May 17, 2026

A Distributional View for Visual Mechanistic Interpretability: KL-Minimal Soft-Constraint Principle

Most current paradigms in visual mechanistic interpretability (MI) remain confined to interpreting internal units of the vision model via heuristic methods (e.g., top-

K

activation retrieval or optimization with regularization). In this work, we establish a theoretical distributional view for visual MI, which models the influence of a feature activation on the natural image distribution, thereby formulating a Kullback-Leibler (KL)-minimal optimization problem to model the MI task. Under this framework, statistical biases are identified within previous MI paradigms, which reveal that they may either be perceptually uninterpretable to humans (i.e., deviate from the natural image distribution), or mechanistically unfaithful to the vision models (i.e., unable to activate model features). To resolve the biases under the distributional view, we propose a model with a KL-minimal soft-constraint principle for visual MI that theoretically balances interpretability and faithfulness. We realize this principle via energy-guided diffusion posterior sampling. Extensive experiments validate the theoretical soundness of the proposed distributional view and demonstrate the practical effectiveness of our paradigm on the DINOv3 vision model.

Guancheng Zhou, Yisi Luo, Zhengfu He +5

School of Mathematics and Statistics, Xi’an Jiaotong University · Shanghai Innovation Institute · OpenMOSS Team +2

Mechanistic Interpretability

May 17, 2026

Beyond Linear Superposition: Discovering Climate Features in AI Weather Models with KAN-SAE

Deep learning weather prediction models achieve remarkable predictive skill yet remain largely opaque: we know little about how they represent physical climate phenomena internally. Mechanistic interpretability through Sparse Autoencoders (SAEs) offers a principled route to decomposing these representations, but existing SAEs assume strictly linear feature superposition - a constraint ill-suited for the highly nonlinear atmospheric dynamics encoded in modern transformers. We introduce KAN-SAE, a sparse autoencoder whose encoder replaces the standard ReLU with learnable per-feature B-spline activations drawn from Kolmogorov-Arnold Networks (KANs), allowing each latent dimension to develop its own nonlinear gating profile. Applied to Sonny, KAN-SAE discovers 975 alive features (vs. 566 for a linear baseline, a 72% improvement) with 20% lower inter-feature redundancy and comparable reconstruction fidelity. Without any climate supervision, KAN-SAE identifies an interpretable European heatwave feature spatially concentrated over western Europe, and a western Pacific typhoon tracker confirmed by causal steering experiments. Our results demonstrate that nonlinear activations are essential for mechanistic interpretability of deep learning weather prediction models, recovering climate features that remain invisible to linear baselines.

Minjong Cheon

Department of Computer Science and Engineering Sejong University

Transformer Interpretability Climate Science Sparse Autoencoders Mechanistic Interpretability Climate Modeling Kolmogorov-Arnold Networks+1

May 17, 2026

Event-Grounded Sparse Autoencoders for Vision-Language-Action Policies

Vision-Language-Action (VLA) policies translate language and visual inputs into robot actions, where their hidden representations directly shape closed-loop behavior. However, mechanistic interpretability tools from language and vision-language models do not transfer cleanly to VLAs: outputs are robot actions rather than human-readable tokens, and interventions can only be tested via expensive closed-loop rollouts. We propose an event-grounded interpretability pipeline that anchors SAE feature analysis to behavioral events rather than text contexts. End-effector keyframes are clustered within each task using visual, state, and temporal cues, linking SAE features to behaviorally salient events and, via optional VLM annotations, to semantic context. To our knowledge, our pipeline is among the first to ground SAE-based VLA analysis in closed-loop behavioral events. Across two simulation architectures and a real-robot study, event-grounded ranking yields the strongest causal effects on OpenVLA and transfers to the continuous action chunks of

π_{0.5}

. SAE is a sparse but imperfect intervention basis: usability varies with architecture and intervention site, and aggressive intervention reveals safety and interpretability limits. Overall, event-grounded SAE analysis emerges as a practical starting point for behavior-anchored VLA interpretability, motivating future work on SAE features beyond action-aligned coordinates, finer-grained closed-loop evaluation, and safe interventions for high-stakes VLA deployments. Code is available at https://github.com/xc-j/Event-SAE.

Xinchen Jin, Aditya Chatterjee, Pranav Kumar +1

Department of Computer Science, Purdue University West Lafayette, IN 47907

VLM Interpretability Sparse Autoencoders Mechanistic Interpretability Vision-Language-Action Models

May 15, 2026

Mechanistically Interpretable Neural Encoding Reveals Fine-Grained Functional Selectivity in Human Visual Cortex

A central goal in understanding human vision is to uncover the visual features that drive neuronal activity. A growing body of work has used artificial neural networks as encoding models to predict cortical responses to natural images, revealing the visual content that activates category-selective regions. However, existing approaches are largely correlational and treat the encoder as a black box, leaving open which image features drive each voxel's response. We introduce Mechanistically Interpretable Neural Encoding (MINE), a framework that opens this black box by applying mechanistic-interpretability tools to localize the features within natural images that drive millimeter-scale (voxel-level) activity. MINE predicts each voxel's response using language-aligned image representations, and produces semantically interpretable descriptions of the features critical for the voxel's activation. We further generalize these per-image features into per-voxel functional profiles. To validate the per-image descriptions, we show they are sufficient to generate images that elicit voxel responses matching the responses to the original images, more accurately than images generated from random or low-attribution controls. Moreover, counterfactually inserting or removing the predicted features from images shifts activation in the expected direction, providing causal evidence. Counterfactual editing guided by the per-voxel activation profiles produces even stronger activation shifts, indicating that the profiles faithfully capture each voxel's selectivity. Finally, we apply MINE to well-studied category-selective brain regions, showing it recovers their known categorical preferences while revealing fine-grained unique voxel structure within each region. Overall, our results establish mechanistic interpretability as a path to discover and causally validate fine-grained hypotheses about neural function.

Idan Daniel Grosbard, Mor Geva, Galit Yovel

1Sagol School of Neuroscience · 2Blavatnik School of Computer Science and AI · School of Psychological Sciences

Mechanistic Interpretability Visual Neuroscience Neuroimaging Neural Encoding

May 14, 2026

When Are Two Networks the Same? Tensor Similarity for Mechanistic Interpretability

Mechanistic interpretability aims to break models into meaningful parts; verifying that two such parts implement the same computation is a prerequisite. Existing similarity measures evaluate either empirical behaviour, leaving them blind to out-of-distribution mechanisms, or basis-dependent parameters, meaning they disregard weight-space symmetries. To address these issues for the class of tensor-based models, we introduce a weight-based metric, tensor similarity, that is invariant to such symmetries. This metric captures global functional equivalence and accounts for cross-layer mechanisms using an efficient recursive algorithm. Empirically, tensor similarity tracks functional training dynamics, such as grokking and backdoor insertion, with higher fidelity than existing metrics. This reduces measuring similarity and verifying faithfulness into a solved algebraic problem rather than one of empirical approximation.

ML Nissen Gonzalez, Melwina Albuquerque, Laurence Wroe +3

MARS & University Heidelberg · MARS · MARS & Stanford University +2

Mechanistic Interpretability Weight-Space Analysis

May 14, 2026

The Rate-Distortion-Polysemanticity Tradeoff in SAEs

Sparse Autoencoders (SAEs) that can accurately reconstruct their input (minimizing distortion) by making efficient use of few features (minimizing the rate) often fail to learn monosemantic representations (highly interpretable), limiting their usefulness for mechanistic interpretability. In this paper, we characterise this tension in learning faithful, efficient, and interpretable explanations, introducing the Rate-Distortion-Polysemanticity tradeoff in SAEs. Under toy-modeling assumptions, we theoretically and empirically show that restricting the SAE to be monosemantic necessarily comes with an increase in rate and distortion. Assuming a generative model behind the input observations, we further demonstrate that the degree of polysemanticity of optimal SAEs is determined by the training data distribution, especially by the probability of features to co-occur. Finally, we extend the analysis to real-world settings by deriving necessary conditions that a polysemanticity measure should satisfy when the data-generating process is unknown, and we benchmark existing proxy metrics on SAEs trained on Large Language Models. Taken together, our findings show that polysemanticity is a data problem that should be accounted for when addressing it at the architectural and optimization level.

Tommaso Mencattini, Francesco Montagna, Francesco Locatello

EPFL · Institute of Science and Technology Austria

Rate-Distortion Theory Representation Disentanglement Sparse Autoencoders Mechanistic Interpretability Sparse Coding

May 14, 2026

Exemplar Partitioning for Mechanistic Interpretability

We introduce Exemplar Partitioning (EP), an unsupervised method for constructing interpretable feature dictionaries from large language model activations with

\sim 10^3\times

fewer tokens than comparable sparse autoencoders (SAEs). An EP dictionary is a Voronoi partition of activation space, built by leader-clustering streamed activations within a distance threshold. Each region is anchored by an observed exemplar that serves as both its membership criterion and intervention direction; dictionary size is not prespecified, but determined by the activation geometry at that threshold. Because exemplars are observed rather than learned, dictionaries built from the same data stream are directly comparable across layers, models, and training checkpoints. We characterise EP as an interpretability object via targeted demonstrations of properties newly accessible through this construction, plus one head-to-head benchmark. In Gemma-2-2B, EP dictionary regions are interpretable and support causal interventions: refusal in instruction-tuned Gemma concentrates in a region whose exemplar ablation can collapse held-out refusal. Cross-checkpoint matching between base and instruction-tuned dictionaries separates the directions preserved through finetuning from those introduced by it. EP regions and Gemma Scope SAE features decompose activation space differently but agree on a shared core:

\sim

20% of EP regions match an SAE feature at

F_1 > 0.5

, and EP one-hot probes retain

\sim

97% of raw-activation probe accuracy at

\ell_0 = 1

. Nearest-exemplar distance provides a free out-of-distribution signal at inference. On AxBench latent concept detection at Gemma-2-2B-it L20, EP at

p_1

reaches mean AUROC 0.881, +0.126 over the canonical GemmaScope SAE leaderboard entry and within 0.030 of SAE-A's 0.911, at

\sim 10^3\times

less build compute.

Jessica Rumbelow

Leap Laboratories

Representation Learning LLM Interpretability Mechanistic Interpretability

May 13, 2026

Why Retrieval-Augmented Generation Fails: A Graph Perspective

Retrieval-Augmented Generation (RAG) has become a powerful and widely used approach for improving large language models by grounding generation in retrieved evidence. However, RAG systems still produce incorrect answers in many cases. Why RAG fails despite having access to external information remains poorly understood. We present a model-internal study of retrieval-augmented generation that examines how retrieved evidence influences answer generation. Using circuit tracing, we construct attribution graphs that model the flow of information through transformer layers during decoding. These graphs represent interactions among retrieved context, intermediate model activations, and generated tokens, providing a graph, circuit-level view of how external evidence is integrated into the model's reasoning process across multiple question answering benchmarks, we observe consistent structural differences: correct predictions exhibit deeper reasoning paths, more distributed evidence flow, and a more structured pattern of local connectivity, while failed predictions show shallower, fragmented, and overly concentrated evidence flow. Building on these findings, we develop a graph-based error detection framework that uses attribution-graph topology features. Furthermore, we show that attribution graphs enable targeted interventions. By reinforcing question-constrained evidence grounding, we reshape internal routing so that answer generation remains guided by the question, leading to more effective integration of retrieved information and fewer errors.

Kai Guo, Xinnan Dai, Zhibo Zhang +5

Michigan State University · Massachusetts Institute of Technology

Retrieval-Augmented Generation LLM Grounding Mechanistic Interpretability

May 13, 2026

Mechanistic Interpretability of EEG Foundation Models via Sparse Autoencoders

EEG foundation models achieve state-of-the-art clinical performance, yet the internal computations driving their predictions remain opaque: a barrier to clinical trust. We apply TopK Sparse Autoencoders (SAEs) across three architecturally distinct EEG transformers: SleepFM, REVE, and LaBraM to extract sparse feature dictionaries from their embeddings. By grounding these features in a clinical taxonomy (abnormality, age, sex, and medication), we benchmark monosemanticity and entanglement across architectures. A single hyperparameter procedure, driven by an intrinsic dictionary health audit, transfers robustly across all three architectures. Via concept steering, we introduce a "target vs. off-target" probe area metric to quantify steering selectivity and reveal three operational regimes: selectively steerable, encoded but entangled, and non-encoded. This framework exposes critical representational failures: "wrecking-ball" interventions that collapse global model performance, and clinical entanglements, such as age-pathology confounding, where it is impossible to suppress one concept without corrupting the other. Finally, a spectral decoder maps these interventions back to the amplitude spectrum, translating latent manipulations into physiologically interpretable frequency signatures, such as pathological slow-wave suppression and

α

-band restoration.

William Lehn-Schiøler, Magnus Ruud Kjær, Rahul Thapa +10

BrainCapture, Kongens Lyngby, Denmark · DTU Health Tech, Technical University of Denmark, Kongens Lyngby, Denmark · DTU Compute, Technical University of Denmark, Kongens Lyngby, Denmark +5

Sparse Autoencoders Mechanistic Interpretability EEG Foundation Models

May 13, 2026

Transformers Linearly Represent Highly Structured World Models

Do transformers, when trained on sequential reasoning traces, build internal models of the underlying task? And if so, does the structure of those internal representations mirror the structure of the domain? We train an 8-layer transformer on Sudoku solving traces and perform a mechanistic analysis of its internal computation. We establish two results. First, the model builds a substructure world model: it does not represent the board state cell by cell, as a human analyst would expect, but organizes information around the rows, columns, and boxes that Sudoku's constraints act on. Second, we identify a naked-single circuit: a small set of dedicated neurons in the final MLP layer, each individually detecting when exactly one digit remains possible for a specific cell, and reliably promoting that digit. These findings show that the geometry of an emergent world model is shaped by the constraint algebra of the domain, not its surface presentation, and that the resulting decision circuit is sparse, monosemantic, and fully interpretable. More broadly, they demonstrate that mechanistic interpretability tools can recover an end-to-end algorithmic account of how a transformer solves a combinatorial reasoning task.

Roman Kniazev, Nathanaël Fijalkow

LaBRI, CNRS · University of Bordeaux · France

Circuit Discovery World Models Mechanistic Interpretability Structured Reasoning Constraint Satisfaction Algorithmic Reasoning

May 13, 2026

Descriptive Collision in Sparse Autoencoder Auto-Interpretability: When One Explanation Describes Many Features

Sparse autoencoders (SAEs) are now standard tools for decomposing language model activations into interpretable features, and automated interpretability pipelines routinely assign each feature a short natural-language explanation. Existing critiques of this practice focus on polysemanticity -- one feature with many meanings -- or on whether explanations predict activations. We identify a complementary, structurally distinct problem we call descriptive collision: many distinct SAE features admit the same explanation. Reanalyzing the largest publicly-available dataset of human-annotated SAE features (Marks et al., 2025), comprising 722 annotated features across Gemma 2 2B and Pythia 70M, we find that the mean annotation string is reused across 3.07 features; 82.1% of features share their annotation with at least one other feature; and the single most common annotation string ("plural nouns") labels 101 distinct features spanning 18 layers and four model components. Information-theoretically, the average annotation resolves only 70% of feature identity. We formalize a property called discrimination, prove that current detection-style auto-interpretability scoring is invariant to collision, and propose two complementary corrective metrics -- collision-adjusted detection and discrimination scoring -- that explicitly penalize explanations that fail to distinguish a feature from its neighbors. The collision problem is independent of, and additive with, previously identified failure modes of auto-interpretability; ignoring it inflates reported feature interpretability by a quantity equal to roughly one-third of the bits required to identify a feature.

Jordan F. McCann

Independent Researcher

LLM Interpretability Sparse Autoencoders Mechanistic Interpretability Explainability Evaluation

May 12, 2026

In-Context Learning Operates as Concept Subspace Learning

Regression and Bayesian accounts of in-context learning (ICL) explain how demonstrations can induce predictors, while mechanistic analyses often identify compact activation directions that steer prompted behavior. However, it remains unclear whether structured demonstrations induce low-dimensional concept inference. We study this question through a concept-subspace view of ICL, in which tasks vary only along intrinsic concept coordinates, although inputs are observed in a high-dimensional ambient space. For ridge and least-squares ICL proxies, prediction decomposes exactly into concept-coordinate regression and off-subspace leakage. Under block-diagonal or near-block-diagonal covariance assumptions, the leading estimation and nuisance-sensitivity terms scale with the dimension of the concept subspace, while residual effects are controlled by cross-subspace coupling. This separation gives a mechanistic prediction: recoverable task information should concentrate in a low-dimensional, task-aligned activation subspace. On CounterFact-derived multi-relation prompts with Llama-3-8B, a 68--73-dimensional subspace of the 4096-dimensional residual stream restores 78.8% of the clean--corrupted accuracy gap, whereas patching the complementary subspace restores 0%. Concept swaps redirect predictions toward injected relations, while random and cross-task matched-rank controls are largely ineffective. Additional experiments on Qwen2.5-7B and a controlled cross-lingual rule task show the same qualitative pattern. These results support concept subspaces as compact, task-aligned mediators of recoverable ICL behavior in structured task families, without implying full-circuit recovery.

Wei Tang, Xinyan Jiang, Fakhri Karray +1

Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) · Shanghai Advanced Research Institute, Chinese Academy of Science

Representation Learning Mechanistic Interpretability In-Context Learning Language Modeling

May 12, 2026

OceanCBM: A Concept Bottleneck Model for Mechanistic Interpretability in Ocean Forecasting

Extreme ocean phenomena are challenging not only to predict but to diagnose, as accurate forecasts alone do not reveal the underlying physical drivers. While recent machine learning approaches achieve strong predictive skill, they remain largely opaque and provide limited guarantees of fidelity to ground-truth physics. We introduce OceanCBM, the first concept bottleneck model (CBM) for spatiotemporal prediction and mechanistic interrogation of ocean dynamics. OceanCBM uses mixed supervision to predict mixed layer heat content, a key precursor of marine heatwaves, while routing information through an intermediate layer of prescribed concepts derived from geophysical fluid dynamics and a 'free' concept. This design imposes soft physical structure without over-constraining the model, and the free concept both regularizes concept predictions and captures residual physical processes. Across ensemble initializations, we show that mixed supervision yields consistent mechanistic representations, whereas prediction-only and prescription-only baselines learn highly variable latent structures despite similar predictive performance. OceanCBM achieves interpretable, physically grounded representations without sacrificing skill, explicitly characterizing the interpretability-performance trade-off.

Sanah Suri, Kieran Ringel, Maike Sonnewald

Department of Computer Science University of California, Davis Davis, CA 95616 · University of Washington, Seattle, WA, USA. · NOAA Geophysical Fluid Dynamics Laboratory, Princeton, NJ, USA.

Concept Bottleneck Models Mechanistic Interpretability Physics-Informed ML Spatiotemporal Forecasting

May 12, 2026

On the Interpretability of Whisper Encodings Using Sparse Autoencoders

While deep transformer-based models have advanced rapidly, their internal mechanisms remain largely a mystery. Recent work has prioritized understanding text-based transformer models, leaving ASR systems largely unexplored. In order to address this gap, we examine the internal representations of Whisper's encoder using a sparse autoencoder. We find diverse monosemantic features across linguistic and non-linguistic boundaries, spanning a hierarchy from phonetic to semantic representations, and conduct a causal feature-steering campaign across this hierarchy, including cross-lingual steering. We further find that steering is more reliable for higher-level features than lower-level ones, an asymmetry that may reflect redundant encoding of lower-level information. Altogether, this work demonstrates that Whisper's encoder represents a surprisingly rich hierarchy of linguistic information that extends well beyond what is strictly necessary for transcription.

Dan Pluth, Zachary Nicholas Houghton, Yu Zhou +1

Vail Systems, Inc. · University of Oregon

Transformer Interpretability Mechanistic Interpretability Automatic Speech Recognition Activation Steering

May 12, 2026

Do Language Models Encode Knowledge of Linguistic Constraint Violations?

Large Language Models (LLMs) achieve strong linguistic performance, yet their internal mechanisms for producing these predictions remain unclear. We investigate the hypothesis that LLMs encode representations of linguistic constraint violations within their parameters, which are selectively activated when processing ungrammatical sentences. To test this, we use sparse autoencoders to decompose polysemantic activations into sparse, monosemantic features and recover candidates for violation-related features. We introduce a sensitivity score for identifying features that are preferentially activated on constraint-violated versus well-formed inputs, enabling unsupervised detection of potential violation-specific features. We further propose a conjunctive falsification framework with three criteria evaluated jointly. Overall, the results are negative in two respects: (1) the falsification criteria are not jointly satisfied across linguistic phenomena, and (2) no features are consistently shared across all categories. While some phenomena show partial evidence of selective causal structure, the overall pattern provides limited support for a unified set of grammatical violation detectors in current LMs.

Hardy, Sebastian Padó

IMS, University of Stuttgart, Stuttgart, Germany · Universitas Mikroskil, Medan, Indonesia.

LLM Interpretability Mechanistic Interpretability Language Modeling

May 11, 2026

Dissecting Jet-Tagger Through Mechanistic Interpretability

Mechanistic interpretability seeks to reverse engineer a trained neural network by identifying the minimal subset of internal components. We perform a mechanistic interpretability analysis of the Particle Transformer architecture, trained on the Top Quark Tagging reference dataset, with the goal of identifying the computational circuit responsible for jet classification and characterizing the physical content of its internal representations. Combining zero ablation, path patching with two complementary on-manifold corruption strategies and linear probing of the residual stream, we identify a sparse six-head circuit that recovers the great majority of the full model performance while admitting a clean source-relay-readout interpretation. In this circuit, a single early layer head serves as the primary causal source, a cluster of middle-layer heads acts as relays selectively attending to hard pairwise substructure and a single late-layer head reads out the aggregated signal. Linear probes show that the residual stream is preferentially aligned with the energy correlator basis over the

N

-subjettiness basis. Within the energy correlator basis, the model preferentially encodes 2-prong substructure observables over the 3-prong observables. A per-layer trained probe further reveals that the apparent single step commitment of the model to a classification decision in the first class attention block is in fact a basis rotation, with the discriminating signal already saturating in the particle attention stack. These results demonstrate that mechanistic interpretability methods developed for natural language models can be used for jet physics classifiers and indicate that gradient descent may rediscover physically meaningful aspects of jet tagging without supervision.

Saurabh Rai, Sanmay Ganguly

Department of Physics, Indian Institute of Technology, Kanpur. Uttar-Pradesh 208016, India

Transformer Interpretability Attention Head Analysis Jet Tagging Mechanistic Interpretability High-Energy Physics

May 10, 2026

fmxcoders: Factorized Masked Crosscoders for Cross-Layer Feature Discovery

Many features in pretrained Transformers span multiple layers: they emerge through stages of inference, persist in the residual stream, or are built jointly by parallel MLPs. Crosscoders (namely, sparse dictionaries trained jointly across layers) aim to recover these cross-layer features in a single shared latent space. We show that standard crosscoders largely fail at this purpose. Although their decoder weight norms spread evenly across layers, a functional coherence metric we introduce reveals that each latent's activation is effectively driven by only one or two layers on average. While functionally coherent latents act as human-interpretable concept detectors (e.g., US states and cities), the layer-localized latents that crosscoders predominantly learn collapse onto surface-level patterns such as digit detectors. We trace this failure to two structural limitations: unconstrained cross-layer parameterization and unregularized cross-layer dependence. We address both by introducing fmxcoders, which (i) replace the encoder and decoder with low-rank tensor factorizations that draw every latent's per-layer weights from a shared cross-layer basis, and (ii) apply stochastic layer masking, a denoising regularizer along the layer axis that penalizes latents whose contribution collapses when a single layer is masked. Across GPT2-Small, Pythia-410M, Pythia-1.4B, and Gemma2-2B, fmxcoders lift mean probing F1 by 10-30 points, surpassing per-layer SAE baselines that standard crosscoders fail to reach, reduce reconstruction MSE by 25-50%, and roughly double mean functional coherence. An LLM-as-a-judge evaluation further shows that fmxcoders recover 3-13

\times

more semantically coherent latents than standard crosscoders across all four base LLMs.

Andreas D. Demou, Panagiotis Koromilas, James Oldfield +2

1The Cyprus Institute · University of Athens · 4Archimedes AI/Athena Research Center +2

Transformer Interpretability Sparse Autoencoders Mechanistic Interpretability