LLM Interpretability

LLM: Large Language Model

Follow this topic Ask about this topic

Momentum

57 papers in the last four weeks, up 73% on the four weeks before. 0.6% of all new papers.

Jul 13Week of Sep 28

Latest papers 546

Jul 17, 2026cs.CL

From Plausible to Actionable: A Position on LLM Self-Explanations

Large Language Models (LLMs) can generate natural language explanations that rationalize their own decisions, a phenomenon commonly referred to as self-explanations. Such explanations have emerged as a promising direction for explainable artificial intelligence (XAI), particularly for interpreting LLM behavior. However, while self-explanations often appear plausible, whether they faithfully reflect a model's underlying reasoning process remains an open question. In this opinion paper, we argue that self-explanations can be highly plausible, questionably faithful, and yet highly actionable. From a traditional XAI perspective, we identify the limitations of standard evaluation protocols for LLM-generated self-explanations and propose practical guidelines for assessing their plausibility and faithfulness. Moreover, we argue that evaluation should extend beyond these criteria to actionability, highlighting applications of LLM rationalization capabilities that support informed decision-making and appropriate action across diverse stakeholders.

Elize Herrewijnen, Benedetta Muscato, Gizem Gezici +1

University of Utrecht, Utrecht, Netherlands · National Police Lab AI, Netherlands Police, Driebergen, Netherlands · Scuola Normale Superiore, Pisa, Italy +1

CoT Faithfulness Faithfulness of Language Model Explanations LLM Interpretability Explainability Evaluation

Jul 15, 2026cs.AI

Tracing LLM Behavior to the Training Data with Empirical Next-Token Distributions

In this paper, we study the connection between an LLM's output distribution and the data used to train it. Specifically, we study the degree to which an LLM's next-token distribution agrees with the empirical next-token distribution (ENTD) given the context in the training data. The ENTD is an appealing target because it is the unrestricted global minimizer of the next-token cross entropy loss used for pretraining, as well as an easily interpretable function of the pretraining corpus. We find that for a significant fraction of inputs, the LLM's distribution agrees with the ENTD almost perfectly, and the average agreement increases with model scale and training compute. Nevertheless, there is a long tail of input sequences where the LLM and ENTD differ significantly, and we examine several possible sources of this discrepancy across the transformer architecture, training procedure, and finite-sample noise in the ENTD estimate itself. More broadly, we hope our findings will encourage more work on ``data-centric mechanistic interpretability,'' a complement to standard mechanistic interpretability that opens the black box of how model behaviors arise from the data, rather than how they are encoded in the learned weights.

Zachary Izzo

NEC Labs America

Language Model Pretraining Transformer Interpretability LLM Interpretability Mechanistic Interpretability

Jul 15, 2026cs.AI

AIMO Interpretability Challenge

We propose the AIMO Interpretability Challenge, a competition on distinguishing robust from spurious reasoning in frontier mathematical language models based on the models' internal mechanisms. The challenge is motivated by a central limitation of standard reasoning benchmarks: strong final-answer accuracy does not reveal whether a model relies on stable reasoning mechanisms or exploits brittle reasoning shortcuts. Building on AI Mathematical Olympiad (AIMO) problems and submissions, together with resources from the Fields Model Initiative, the competition will provide (1) newly-published olympiad-level math reasoning problems and their symbolic representations, allowing generation of novel functional variants, (2) access to frontier reasoning models, and (3) assessments of models' adversarial robustness on these problems. Participants will use these resources, along with our computing infrastructure support, to develop methods for identifying which models solve problems robustly. Our competition will also create a new, open robustness benchmark and baseline systems, aiming to provide a lasting foundation for standard benchmarking in mathematical reasoning and interpretability. Scientifically, the competition connects interpretability and generalization research around a central question in AI research: can we determine if, and to what extent, the decision-making of frontier AI models is generalizable and thus, reliable?

Michal Štefánik, Philipp Mondorf, Andreas Waldis +11

National Institute of Informatics, Japan · Munich Center for Machine Learning / MaiNLP LMU · University of Tübingen +6

Mathematical Reasoning Benchmarks LLM Interpretability Mechanistic Interpretability Mathematical Reasoning Adversarial Robustness of Language Models

Jul 14, 2026cs.CL

What Models Express, Suppress, and Resist: Auditing Open-Weight LLMs with Persona Vectors

What a language model will and will not do is largely set during post-training, but which behaviors it expresses, hides, or resists is not revealed by prompting alone. Persona vectors, behavioral directions in activation space, can probe this organization, but prior work covers only a handful of traits. We present the first systematic application of persona vectors at this scale, compiling a 53-trait inventory across four behaviorally distinct domains and labeling every trait in two open-weight models as natural (expressed at baseline), steerable latent but amplifiable, or intractable (resistant to standard extraction). Both models default to helpful, task-oriented behavior: all nine agentic traits are natural, and their default clinician behavior matches a board-certified psychologist's independent desirability judgments on 16 of 17 traits. Steering produces its largest gains on traits these defaults exclude: hyperbole, hallucination, and sycophancy. The same asymmetry holds across all 171 generic-trait pairs: two steerable traits can collapse the composition, but pairs involving a default never do. Where standard extraction fails on a trait like "evil," a vector transferred from a fine-tuned variant still recovers it, with the residual refusals appearing inside the model's chain-of-thought. Persona vectors are most informative not as a set of controls but as a probe of behavioral organization.

Winston Zeng, Ali Emami, Jinho Choi

Emory University

Language Model Steering LLM Interpretability Personality Modeling in Language Models Language Model Auditing

Jul 13, 2026cs.LG

Inside the Unfair Judge: A Mechanistic Interpretability Account of LLM-as-Judge Bias

Existing studies of LLM-as-judge scoring bias work predominantly at the input-output level: they perturb inputs, measure score deltas, and propose prompt-level mitigations. We argue that the same biases admit a representation-level account in the judge's hidden state, complementary to the input-output view and operationally useful in ways it does not afford. We report three findings, across seven judges, seven bias types, and nine benchmarks. Geometry: baseline judging inputs occupy a tight activation manifold while biased inputs are displaced along a low-dimensional, type-specific subspace that sharpens with depth and is recovered consistently by three families of estimators. Causal control: steering hidden states along this subspace drives scoring in both directions, forward shifts reproducing biased scoring on clean inputs and reverse shifts restoring baseline scoring on biased ones, while matched-norm random directions produce shifts an order of magnitude smaller. Operational: a simple linear projection onto the same bias-direction features anticipates judge failures on three entirely unseen benchmarks, substantially outperforming text-based alternatives. Reading bias as activation geometry, rather than as input-output noise, unifies geometric structure, causal control, and operational prediction within a single framework. The project page is available at https://xzx34.github.io/unfair-judge/

Zixiang Xu, Sixian Li, Huaxing Liu +4

AMAP, Alibaba Group · Mohamed bin Zayed University of Artificial Intelligence · University of Southern California +1

LLM-as-a-Judge LLM Interpretability Language Model Bias Evaluation Bias in LLM-as-a-Judge Representation Geometry in Language Models

Jul 12, 2026cs.LG

Weight-Adjusted Gradients Reveal Parameter Importance and Failure Modes in LLMs

Understanding which parameters are influential in Large Language Models (LLMs) is central to improving their efficiency, reliability, and interpretability. We introduce Weight-Adjusted Gradients (WAG), a simple yet effective approach for estimating parameter importance that explicitly captures the interaction between model weights and first-order gradient information and identifies parameters that disproportionately influence model behavior, such as those responsible for collapse phenomena in LLMs. Across a range of models and settings, we show that WAG surfaces a tiny but critical subset of parameters (< 0.5 parts per million or 0.00005% of model size) whose modification leads to dramatic degradation in performance, indicating a novel failure mode. These findings also reveal a previously underexplored interplay between weights and gradients, suggesting that parameter importance cannot be fully understood through either signal alone. We demonstrate the practical utility of WAG across several diverse applications, such as expert allocation in Mixture-of-Experts (MoE) architectures, targeted unlearning, mixed-precision quantization, and layer selection for knowledge editing. In sum, WAG can serve as a unified approach for analyzing, debugging, and controlling LLMs, and opens new directions for principled parameter-level interpretation.

Shrestha Datta, Hongfu Liu, Anshuman Chhabra

University of South Florida, Tampa, FL, USA · Brandeis University, Waltham, MA, USA

Gradient-Based Attribution LLM Interpretability LLM Reliability

Jul 12, 2026cs.AI

Laguerre Geometry for Interpreting Large Language Models

Existing hypotheses represent a concept in an LLM as a single point, a linear direction, or a Gaussian cluster, yet it remains unclear how and why such structures emerge. Here, we show that concept geometry can be precisely characterized via Laguerre Geometry, in which a concept is defined as a region--a Laguerre-Voronoi cell or a union of cells--allowing us to strictly define, measure, and separate concepts. Building on this formulation, we show that finer-grained concept structures, such as inclusion and hierarchy, are naturally revealed by the Laguerre weights. We then push this geometry inside the transformer. Decomposing each layer into piecewise-linear operators, we show that a token's hidden trajectory is governed by two coupled mechanisms: a static tree of self-contained piecewise-linear flow, and a dynamic transport that hops the trajectory across trees when cross-token attention fires. This decomposition yields Geometric Lens, a training-free, hyperparameter-free method for reading out the exact concept a hidden vector encodes at any layer. We also develop Laguerre Autoencoder, a 2D visualizer that renders both the decision geometry and a model's full reasoning trajectory in one view. Finally, we move beyond explanatory geometry toward actionable interpretability, showing that Geometric Lens recovers the correct factual token when a model is prompted with in-context interference. The code is available on GitHub.

Chunwei Ma, Russell Wolfinger

JMP Statistical Discovery, Cary, NC, USA

Transformer Interpretability LLM Interpretability Representation Geometry in Language Models

Jul 11, 2026cs.AI

SPARK: Susceptibility-Guided Profiling and Steering of Latent Reasoning States in Large Language Models

Reasoning failures in large language models (LLMs) are usually evaluated from final answers, but a wrong answer does not reveal why the model failed. The same incorrect output may reflect missing capability, an unstable reasoning trajectory, or a failure to activate a reasoning state that is already available in the frozen model. Existing prompting and benchmark-based evaluation methods mostly operate at the output level, while generic activation-steering methods typically apply global directions without diagnosing which examples require intervention. In this paper, we introduce SPARK, which uses hidden-state response to diagnose whether a model internally enters an effective reasoning state and to guide lightweight test-time steering. The key observation is that raw hidden-state susceptibility is strongly confounded by prompt length, especially in programmatic and algorithmic reasoning where harder serialized instances naturally become longer. SPARK therefore uses length-controlled susceptibility to separate input-scale effects from residual reasoning activation, and combines this signal with cross-layer coordination to select reasoning-active anchors and under-activated hard examples. We use FRONTIER-4.5K as a controlled programmatic reasoning suite for latent profiling and difficulty-aware analysis, and evaluate SPARK-Steering on GSM8K and MATH-500 with forward-only benchmark profiling. Our method improves Qwen3 series models consistently; on MATH-500, accuracy rises from 82.0% to 84.6% for Qwen3-4B and from 82.4% to 85.6% for Qwen3-8B. These results suggest that susceptibility can serve not only as a diagnostic signal for reasoning failures, but also as a practical guide for targeted test-time intervention.

Dongxu Zhang, Yiding Sun, Zihao Guo +5

Xi’an Jiaotong University · Peking University · Tencent

LLM Interpretability Activation Steering Language Model Probing LLM Reasoning

Jul 11, 2026cs.CL

Belief-reality separation lives in routing over a shared value slot in language models

Capable language models hold what a character believes apart from what is true: told "Anna believes the cup is blue; in reality it is red," they answer blue about Anna and red about the world. Where in the computation does that separation live? We show it rests on two separable mechanisms at two positions. A generic value slot binds the attributed value. A router at the query position selects which frame, the character's belief or reality, a query reads out. Two routes fill the slot: an asserted belief, whose value the text supplies, binds in directly; a derived belief, whose value must be inferred from what the character could see, arrives by a visibility-gated lookback. A subspace trained on either route steers the other, and only the derived route depends on described visibility. The slot itself carries no belief-reality tag: intervening on it moves a reality readout as strongly as a belief one. The separation lives instead in a dissociated pair of routing subspaces, which flip a query between frames without injecting the donor's value. These results hold across three architectures, on stimuli de-confounded against theory-of-mind-benchmark shortcuts; the behavior itself emerges between 3B and 7B across five model families. This paper develops the single belief-reality axis in depth; a companion paper shows the same slot-and-router format is shared across the other non-actual contexts a sentence can open (counterfactual, fictional, temporal).

Oliver Steele, Jiangtao Wen, Yuxing Han

Tsinghua University · NYU Shanghai

Theory of Mind LLM Interpretability Circuits in Language Models Representation Geometry in Language Models

Jul 11, 2026cs.CL

One mechanism for many mental spaces: a shared router over a value slot in language models

Language builds discourse contexts other than the actual: a painting, a belief, a memory, a hypothetical. Each is a mental space in which the same entity can take a different value, as when a flower is red in reality but purple in a portrait. Formal semantics keeps these contexts apart because their logics differ (modal, temporal, doxastic, depictive); Fauconnier's mental-space theory treats them as one space-building operation. We ask which of these a transformer language model implements, and find a mechanistic version of Fauconnier's unification. The model uses one router/slot format across the inventory: a reusable value slot stores attributed content, and a causally manipulable router (the space index) selects which space is read. A subspace trained with Distributed Alignment Search to control one space type, counterfactual, belief, fictional, or temporal, also controls the others, well above a random floor, on three model families; belief, which formal semantics marks as a distinct case, is not specially separated. The router is low-rank, composes additively with entity identity, and acts through a few late-layer heads. Two further results show the mechanism drives inference and composes: a subspace trained on a rule-derived conclusion flips what the model infers while dissociating from what it reports, and composing space-builders mints a fresh router over the shared slot. This paper establishes the cross-type generality. A companion paper develops belief in depth, because of its special status in philosophy, psychology, and linguistics (epistemology, theory of mind, and propositional attitude reports).

Oliver Steele, Jiangtao Wen, Yuxing Han

Tsinghua University · NYU Shanghai

LLM Interpretability Circuits in Language Models Representation Geometry in Language Models

Jul 10, 2026cs.LG

All Explanations are Wrong, But Many Are Useful: Exploring the Rashomon Explanation Set with Large Language Models

Explaining machine-learning models is increasingly important for decision-making and consumer trust, yet it is widely believed to come at a cost: existing Explainable AI (XAI) methods suffer from a persistent accuracy-explainability trade-off. We argue that this trade-off is not fundamental, but an artifact of treating explanation and prediction as separate objectives; when properly coupled, they become complementary, so that equipping a model to explain itself improves, rather than degrades, its accuracy. We introduce the Rashomon Explanation paradigm, which builds a set of faithful, prediction-guiding explanations rather than a single one, and prove that this set is generally non-empty and that explanation fidelity bounds the performance of the models it guides. To explore this set, we propose RashomonLLM, an Explanation-Prediction-Reflection agentic workflow that generates explanations in natural language by iteratively aligning them with predictions, and we prove it converges and recovers the full set. Across customer-churn classification, clinical survival regression, and industrial click-through prediction on large-scale live-streaming logs, RashomonLLM significantly outperforms state-of-the-art prediction and XAI baselines on both accuracy and explanation quality, with gains driven by explanation fidelity and robust to distribution shifts, temporal splits, and seeds. Our framework thus advances business performance while laying the groundwork for consumer trust.

Pan Li

Scheller College of Business, Georgia Tech.

Explainable Artificial Intelligence LLM Interpretability

Jul 9, 2026cs.CL

TypeProbe: Recovering Type Representations from Hidden States of Pre-trained Code Models

State-of-the-art code models achieve impressive performance, yet the extent to which they internally encode type information remains poorly understood. We probe the residual streams of pretrained code models for internal type representations using a parallel dataset of Java and Python code examples. Our results show that cross-lingual type representations emerge even from untyped code. Moreover, we test whether hidden states linearly encode the result type implied by typed function application by training probes on one language to infer argument and result types in the other. Finally, we find that this structure is partly robust to lexical perturbations and cross-language syntactic variations. To the best of our knowledge, prior work on interpretability of code models has not directly targeted formal type semantics or cross-lingual type representations. We release our code and datasets.

Giuliano Gorgone, Fausto Carcassi

ILLC, University of Amsterdam, Amsterdam, The Netherlands

LLM Interpretability Representation Geometry in Language Models Code Language Models Language Model Probing

Jul 8, 2026cs.CR

Mechanistic Interpretability of LLM Jailbreaks via Internal Attribution Graphs

Large language models (LLMs) exhibit remarkable capabilities but remain highly vulnerable to adversarial prompts and jailbreak attacks. Existing approaches primarily analyze these failures through input-output behaviors or attribution methods, offering limited insight into how adversarial perturbations alter the model's internal reasoning. Consequently, the mechanisms underlying unsafe or incorrect behaviors remain poorly understood. We introduce a mechanistic framework for diagnosing LLM vulnerabilities using paired internal computation graphs, which represent prompt-specific inference as structured causal interactions among latent features. By constructing and aligning computation graphs for clean and attacked prompts, we reveal that adversarial attacks induce systematic transformations of internal reasoning, including suppression of safety-relevant components, emergence of attack-specific features, and rerouting of computation paths. Building on this representation, we propose a unified framework that (i) decomposes computation into invariant, suppressed, and emergent structures, (ii) identifies recurring vulnerability motifs associated with failure modes, and (iii) performs causal interventions on nodes, paths, and subgraphs to directly evaluate their contributions to attack success. This enables a transition from descriptive attribution to causal diagnosis of model failures. Experiments across multiple open-source LLMs and diverse adversarial and jailbreak benchmarks demonstrate that structural deviations in internal computation graphs strongly correlate with unsafe behaviors. Furthermore, targeted interventions on identified vulnerability motifs improve model robustness, establishing internal computation graphs as a principled foundation for understanding, diagnosing, and mitigating LLM vulnerabilities.

Anupam Wagle, Ifrat Ikhtear Uddin, Chaowei Zhang +1

Department of Computer Science, University of South Dakota, USA · School of Information and Artificial Intelligence, Yangzhou University, Yangzhou, China

LLM Interpretability Adversarial Attacks on LLMs Mechanistic Interpretability LLM Jailbreak Attacks Adversarial Robustness of Language Models Causal Interventions in Language Models

Jul 8, 2026cs.LG

Distributed Sparse Interventions in Language Models

Language models perform a wide range of tasks at varying levels of abstraction with the capacity to flexibly infer tasks from context, execute multiple tasks simultaneously, and select among competing tasks. To study the role of model components in task behaviour, their causal influence can be investigated through interventions. Prior work on model steering has largely focused on interventions along global directions in activation space, modeling task representations as approximately linear and additive. By studying interventions at the neuron level, we find substantial, neuron-specific nonlinear effects on model outputs that are not captured by current steering approaches. We introduce Distributed Sparse Interventions (DSI), an intervention approach that considers nonlinearities and interactions between neurons across layers to identify sparse sets of neurons that elicit task-relevant computations. Across a range of tasks, we demonstrate that DSI can activate task behaviour in instruction-tuned language models by localising and intervening on as few as 0.01% of neurons, highlighting the effectiveness of sparse, distributed interventions in the neuron basis. Additionally, adopting a set-based perspective enables computations over the identified neuron sets, offering insights into the roles of individual neurons by analysing their effects across tasks. Through sparse interventions, DSI enables fine-grained control over model behaviour, localisation of task-relevant neuron sets, and furthers our understanding of task composition.

Maximilian S. Ernst, Lorenz Linhardt, Aaron Peikert +1

Max Planck School of Cognition · Center for Lifespan Psychology Max Planck Institute for Human Development · Machine Learning Group Technische Universität Berlin +5

Language Model Steering LLM Interpretability Circuits in Language Models Causal Interventions in Language Models

Jul 8, 2026cs.LG

Dissociating the Internal Representations of Sycophancy in LLMs

Large Language Models (LLMs) frequently exhibit sycophancy, where they agree with a user's statement even when incorrect. While sycophancy is often treated as a single defined behavior, it can manifest in substantially distinct ways and circumstances, raising the question of whether this multi-faceted nature is reflected in its internal mechanisms. To address this gap, we dissociate the representations of sycophancy into factual and opinion subtypes -- motivated by the distinction between verifiable claims and subjective beliefs. We train linear probes and construct steering vectors on activations of one subtype and evaluate their transfer to the other subtype to measure to what extent they share representations. We find evidence that different LLMs represent these subtypes differently, with either more unified or more distinct and causally interfering representations. This method of dissociation offers a promising framework for studying the representational structure of complex model behaviors.

Anthony Baez, Sheer Karny, Pat Pataranutaporn

Media Lab, MIT, Massachusetts, USA.

LLM Sycophancy LLM Interpretability Representation Geometry in Language Models

Jul 7, 2026cs.LG

Reward Valuation in Large Language Models: Causal Induction of Anhedonia

Recent frontier models mimic complex aspects of human cognition. Here we ask whether this alignment extends into reward valuation, which we assess in a mechanistic framework. Specifically, we use clinical tests that were developed to evaluate anhedonia in human subjects with major depressive disorders. Mechanistically, anhedonia is frequently associated with dysregulation in the Nucleus Accumbens (NAc) and the broader dopaminergic reward system. While neuroimaging has localized these deficits, establishing a causal link between NAc activity and specific behavioral symptoms remains a challenge. We use these ideas from neuroscience to functionally identify reward-anticipatory units in state-of-the-art AI models, and evaluate their causal involvement via targeted perturbations. We find that not only are such model units predictive of NAc brain recordings, their perturbation also induces behavioral effects mirroring human anhedonia: the model opts for low-effort, low-reward tasks in effort-based decision-making paradigms. Crucially, our results demonstrate that this represents a specific deficit in self-centered reward valuation and anticipation--rather than a loss of task capability, reward calculation, or effort avoidance. This induced vulnerability aligns with clinical measures of anhedonia and motivation in humans, such as DARS and MAP-SR, instruments that contain no reward-related vocabulary, ruling out a purely lexical account of the perturbation effect. Taken together, our results suggest reward valuation circuits in AI models that functionally mimic those in humans.

Melika Honarmand, Samin Mahdipour Aghabagher, Martin Schrimpf

NeuroAI Laboratory, EPFL

Reward Modeling Cognitive Modeling LLM Interpretability Causal Interventions in Language Models

Jul 6, 2026cs.CL

How Much is Left? LLMs Linearly Encode Their Remaining Output Length

Large language models generate one token at a time, yet their responses show remarkably consistent length structure: step-by-step solutions converge in predictable token counts, retrievals stop after a few sentences, retractions extend responses by measurable amounts. We ask whether the model carries an internal estimate of how much response remains. Training minimal-capacity linear probes on frozen hidden states of three open-weight 7-8B models across seven completion-style datasets, we find three converging pieces of evidence. First, total response length is linearly decodable from the prompt's last hidden state alone, before any output is emitted. Second, probe directions trained on natural-language datasets transfer broadly, including to controlled synthetic completions never seen in training, outperforming a statistical baseline; the converse direction generally fails, and this asymmetry is itself informative. Third, on curated high-loss completions, the probe's per-position estimate shifts upward at the moment the model retracts and restarts a partial solution, a directional behavior no position-only predictor can reproduce (qualitative, not aggregate). We frame this as approximate estimation of remaining generation length, distinct from exact-counting impossibility results for transformers, and interpret it as evidence that LLMs maintain a plan-like internal representation of output length (decodable, not necessarily used causally).

Mohamed Amine Merzouk, Dmitri Carpov, Mirko Bronzi +2

Mila, Quebec AI Institute · McGill University · LawZero

LLM Interpretability

Jul 6, 2026cs.LG

Latent Programming Horizons in Coding Agents

A coding agent solving a software-engineering task spends dozens of steps reasoning, editing code, and running tests, yet little is known about what the underlying language model internally represents about the program it is working on. We show that the residual streams of language models under coding agents linearly encode properties of the evolving program: a logistic-regression probe on hidden states is able to decode whether the current code parses, passes its test suite, reduces the number of failing tests, and introduces regressions, reaching AUC up to 0.83 for correctness across two models and two benchmarks. Our second finding is more surprising: these representations run ahead of the agent's own edits. Probes trained to predict the outcome of future edits (before they are materialized and written on disk) achieve performance above chance up to roughly 25 steps in advance. We call this the agent's latent programming horizon. As a proof of external validity, we show that the probes transfer across benchmarks without retraining. Our positive results open calls for more research in mechanistic interpretability of coding agents.

André Silva, Han Tu, Martin Monperrus

KTH Royal Institute of Technology Stockholm, Sweden

LLM Interpretability Coding Agents Language Model Probing

Jul 6, 2026cs.CL

Wrong Before Right: Late Rescue and Interface Failure in Aligned Language Models

We study how correctness is assembled inside aligned language models, not only whether the final answer is right. Using layer-wise difference-in-differences (DiD) trajectories over polarity-controlled minimal pairs, we identify the wrong-dip: in mid layers (25-90% depth), internal preference transiently commits to the incorrect answer and is rescued only by late-layer correction. We verify this causally with patchscope-style activation transplantation across 17 models, three families, and 64x scale (0.5B-32B). Four findings follow. (1) Alignment amplification of the causal wrong-dip is recipe-specific and emergent: it emerges at 3B in Qwen2.5, remains high, and peaks at 32B (paired t up to 9.7), reverses in Llama-3-8B (t=-2.31), and sits between for Mistral-7B. (2) The dip predicts real compression failures: high-dip items are 3-7x more likely to flip under late-layer low-rank compression, block dropping, or structured pruning, while quantization flips are dip-blind, a double dissociation confirmed by late-layer ablation. (3) The dip is trainable: a LoRA fine-tune with a mid-layer wrong-margin penalty matches output-only SFT accuracy while cutting the causal dip by 67-70% and improving compression robustness; output-only SFT worsens the causal dip by up to 2.8x at perfect surface accuracy. (4) With controlled readouts, the phenomenon survives natural-language I/O: dip stratification of structural-damage failures is significant on naturalistic vignettes, and free-form fragility separates into a dip-auditable late-rescue layer and a dip-blind interface layer. Together, output-level correctness can hide a late-rescue process that governs compression risk, post-training quality, and evaluation distortion.

Jiaqi Deng

Independent Researcher

LLM Alignment LLM Interpretability Causal Interventions in Language Models

Jul 5, 2026cs.AI

Unsupervised Features Mining via Activation Geometry

Interpretability methods aim to reveal the features represented inside large language models (LLMs). Many existing methods begin with labeled examples of a human-defined concept that may reflect human biases, and then identify how that concept is represented within the model, for example in its activation space or through other decomposition methods. We introduce \emph{Mining via Activation Geometry} (MAG), a simple unsupervised framework for extracting reasoning features from model activations by prepending the same natural-language instruction

Q

to every input

p

, where

Q

defines the reasoning feature of interest, such as Can this object be found in the desert?'' or Is this prompt malicious?'' We measure how the instruction changes the model's internal representation using

m(Q \mid p) - m(p)

at a single readout point. We explore eight different MAGs. The extracted reasoning features predict the models' own world understanding and judgment, can be approximated into a single activation direction, we found that some features are more linearly represented and some less, this linear representation, which is vector steering, can change the LLMs' decisions through activation steering by injecting reasoning features. Finally, we use the same method to select the best training datasets for prompt-injection classifier probes: while similarity between ordinary activations is almost unrelated to downstream performance, RFD-based similarity achieves

94.7\%

Top-1 and

100\%

Top-2 accuracy.

Amit LeVi, Elad David, Max Fomin

Zenity Tel Aviv, Israel

LLM Interpretability Unsupervised Representation Learning Representation Geometry in Language Models Activation Steering Language Model Probing

Jul 4, 2026cs.CL

Can Dialects Be Steered Like Languages? Sparse Neurons and Distributed Directions in Arabic LLMs

Dialectal data are scarce relative to Modern Standard Arabic (MSA), causing Arabic LLMs to overproduce MSA and struggle with dialectally accurate generation. This raises a fundamental interpretability question about where and how dialectal features are encoded within model internals and whether these representations can improve dialect generation without fine-tuning. We study two inference-time approaches as interpretability probes and control mechanisms. First, neuron-level analysis identifies sparse populations that encode dialect-specific features and tests whether amplifying or suppressing them steers model outputs toward target dialects. Second, vector steering extracts dialect-specific activation directions and injects them during inference, motivated by feature entanglement at the neuron level. We find that these neurons are real but only partially explanatory. They occupy under 1% of MLP dimensions but span only 5% to 21% of the residual dialect direction. This limited coverage is causally consequential. Neuron steering reinforces dialect in some varieties when the prompt is already dialectal but cannot induce it from MSA prompts, whereas vector steering succeeds in both settings. Arabic dialects are therefore steerable mainly through distributed rather than localized representations

Kareem Elozeiri, Mervat Abassy, Omar Kallas +4

1Mohamed bin Zayed University of Artificial Intelligence · 2Qatar Computing Research Institute, Hamad Bin Khalifa University · 3Tohoku University +1

Arabic NLP LLM Interpretability Causal Interventions in Language Models Representation Geometry in Language Models Activation Steering Text Generation

Jul 3, 2026cs.CL

Reading Between the Dots: Decoding Hidden Computation across Filler Tokens

Frontier LLMs can perform multi-step reasoning over content-free filler tokens like dots or counting sequences, producing correct answers with no visible chain-of-thought (CoT). This is a limit case for behavioral oversight, where surface tokens carry no information about the underlying reasoning. But hidden from the output is not the same as hidden from us. On four task families (fact retrieval, parallel numeric composition, string manipulation, and in-context computation), two open-weights frontier models (DeepSeek V3, Kimi K2) compute over filler tokens in a legible way: attention routes the question through the filler region to the answer, logit-lens readouts show retrieved facts emerging early and their composition crystallizing in late layers, and KV-cache transplants at filler positions causally swap outputs between examples. We introduce an unsupervised decoding pipeline that takes only hidden states as input and recovers intermediate values with 82-94% accuracy (best LLM judge) across both models and all four tasks, without ground-truth labels or training. Even without a judge, the hidden values are already directly in the pipeline's top-2 tokens 35-85% of the time. The uplift persists whether the filler is prefilled or the model generates the filler itself. On these cleanly decomposable tasks, hidden computation that defeats behavioral CoT monitoring is readable from the residual stream, which suggests that monitorability is a property of the model's full computational trace rather than only its surface tokens.

Kaley Brauer, Claudio Mayrink Verdun, Samuel Marks

Harvard University · 4Anthropic

LLM Auditing LLM Interpretability Implicit Reasoning in Language Models Language Model Probing LLM Reasoning Reasoning Trace Analysis

Jul 2, 2026cs.AI

Fast Multi-dimensional Refusal Subspaces via RFM-AGOP

Steering and monitoring activations in Large Language Models (LLMs) are increasingly used for both safety and interpretability. Early work assumed behaviours are encoded along single linear directions, but recent findings suggest complex behaviours, such as the refusal to answer harmful queries, live in multi-dimensional subspaces. However, existing methods for extracting these subspaces are computationally expensive, which becomes prohibitive on reasoning models who produce long reasoning traces. By adapting the Recursive Feature Machine (RFM) algorithm -- which can be computed efficiently -- with a probe-informed initialization, we are able to identify the multi-dimensional refusal subspace in seconds, on reasoning (Qwen 3) and non-reasoning (Qwen 2.5) models. While RFM allows for faster subspace identification, it also showed better performances on the ablation task than its alternatives. More work is planned to better understand the relations between subspaces found by different methods. If confirmed, RFM could be a cheap and scalable complement to existing subspace-extraction methods in LLMs.

Thomas Winninger

Télécom SudParis, Évry-Courcouronnes, France · ENS Paris-Saclay, Gif-sur-Yvette, France

LLM Interpretability LLM Safety Refusal Behavior in Language Models

Jul 1, 2026cs.CL

Conversable Complexity: Agentic LLM Collectives as Interpretable Substrates

Complexity and interpretability rarely coincide: systems rich enough for complex behaviours to emerge are usually too opaque to question, while transparent ones are too simple for anything complex to emerge. A single large language model (LLM) is a static artefact, hardly exhibiting any of the emergent properties we associate with life. This changes through interaction: populations of LLMs display emergent dynamics absent from isolated models. Furthermore, LLMs can be endowed with persistent memory, tools and shared skills, and the capacity to initiate actions unprompted, i.e., turning LLMs agentic. In this paper, we argue that such collectives of agents can serve as a computational substrate for Artificial Life (ALife) research. Critically, since the agents communicate in natural language, their collective behaviour can be directly interrogated by examining textual traces and asking the agents themselves. We outline the notion of interpretability in language-model research and extend it for collectives of agents. Lastly, we survey recent examples of agentic LLM collectives that already instantiate the idea of agentic substrates, from controlled experiments to deployments in the wild.

Elias Najarro, Ane Espeseth, Eleni Nisioti +2

IT University of Copenhagen, Denmark · University of Oslo, Norway · Sakana AI, Japan +1

Artificial Life LLM Interpretability Agent-Based Modeling Multi-Agent Systems Multi-Agent LLMs Emergent Behavior in Multi-Agent Systems

Jul 1, 2026cs.CL

Understanding Large Language Models

Large Language Models (LLMs) represent one of the most significant advances in AI and natural language processing in recent years. Still, many pressing questions about their mechanisms, capabilities, and relationship to human cognition remain highly debated. This chapter aims to outline our current understanding of LLMs by discussing recent evidence on emerging capabilities and their mechanistic implementation within processing layers. We begin with a concise overview of the Transformer architecture, emphasizing how the attention mechanism enables training on massive datasets, allowing LLMs to function as generalist rather than specialized models. Next, we examine emergent LLM capabilities that appear to resemble aspects of human cognition, including symbolic reasoning, theory of mind, and deception strategies. Several studies provide evidence that LLMs can solve tasks previously thought to require human-like cognition. Other studies reveal insightful failure cases that shed light on the differences between human and LLM cognition. Alongside these findings, we review explainable AI approaches ranging from neuron activation analysis to circuit tracing. In the final section, we address current debates concerning what LLMs genuinely understand versus what they merely appear to understand. Prominent arguments against AI anthropomorphism point to the simplicity of LLM training objectives, claiming that LLM behavior is better explained by pattern memorization of training data than by genuine cognition. We argue that this standpoint is guided by misconceptions about optimization processes and cognitive capacity, and advocate for a more nuanced discussion of LLM cognition that neither dismisses the differences between humans and LLMs nor precludes the possibility of AI cognition through overly simplistic reductionist arguments.

Yannik Keller, Thomas Eisenmann

LLM Interpretability

Jul 1, 2026cs.CL

Logit-Contribution Scoring Identifies Non-Literal Retrieval Heads

In long-context use, large language models frequently synthesize answers from the meaning of a relevant context span rather than literally copy-pasting them. Identifying which attention heads perform this synthesis matters for interpreting long-context model behavior. Yet existing detectors miss these heads by construction: they reward heads whose attended token matches the generated token, a literal-copy criterion that captures where a head reads but not what it writes through its output-value (OV) circuit, the very mechanism that carries non-literal retrieval. We introduce Logit-Contribution Scoring (LOCOS), a write-aware detector that scores each head by the projection of its OV-circuit output onto the answer-token unembedding direction, contrasting needle and off-needle source positions in a single forward pass. Across three model families (Qwen3, Gemma-3, OLMo-3.1), mean-ablating the top LOCOS heads on the NoLiMa non-literal retrieval benchmark collapses ROUGE-L at lower head counts than prior attention-based detections; on Qwen3-8B, ablating 50 heads drives ROUGE-L from 0.401 to 0.000 while the strongest baseline still retains 0.292. The selected heads are retrieval-specific: parametric recall and arithmetic reasoning stay at baseline under the same ablation. On Qwen3-8B, the same ablation also drops MuSiQue from 0.55 to 0.08 and BABI-Long from 0.62 to 0.20, while a random-heads control stays within 0.05 of baseline.

Aryo Pradipta Gema, Beatrice Alex, Pasquale Minervini

QUniversity of Edinburgh · KHeriot-Watt University · V Miniml.AI

Long-Context Retrieval Attention Head Analysis Long-Context Language Model Evaluation Long-Context Language Model Inference LLM Interpretability Multi-Head Attention

Jul 1, 2026q-fin.CP

Shapley in Context: Explaining Financial Language with Domain Expertise

In recent years, large language models have achieved remarkable success and have seen growing adoption in financial applications. At the same time, explainability remains critical in finance, a domain characterized by high stakes and strict regulatory requirements. Although numerous methods have been proposed to explain black box machine learning models, the majority of these approaches are designed for general purpose tasks and do not incorporate domain specific knowledge. In this work, we study the explainability of financial textual data modeled by large language models through the lens of the Shapley value. Specifically, we investigate whether Shapley based attributions align with established financial domain knowledge. Through rigorous theoretical analysis and extensive empirical evaluations, we demonstrate that Shapley values can yield explanations that are consistent with financial reasoning and can offer meaningful insights into the model's behavior in text based financial applications.

Dangxing Chen, Pengzhan Guo

Zu Chongzhi Center, Duke Kunshan University, Kunshan, Jiangsu, CN · Digital Innovation Research Center, Duke Kunshan University, Kunshan, Jiangsu, CN

LLM Interpretability Shapley Value Attribution Explainability Evaluation

Jul 1, 2026cs.LG

Prototype Language Models

Knowing which training examples drive outputs is fundamental to auditing, correcting, and understanding language models, yet for modern LLMs this remains expensive, approximate, and largely post-hoc. Standard language models generate tokens through a dense network pathway, causing training data's influence to be distributed across parameters rather than organized along explicit, traceable components. We introduce a prototype language model architecture, Prototypes for Interpretable Sequence Modeling (PRISM), that forms each prediction via a sparse, non-negative mixture of learned prototypes, trained with clustering objectives that anchor each prototype to coherent neighborhoods of training examples. Across architectures from 130M to 1.6B parameters trained on up to 50B tokens, prototype language models either surpass or remain within 2.5 percentage points on average downstream accuracy of matched dense baselines. We show that sparse prototype structure localizes curvature in the loss landscape, yielding a more tractable Hessian and enabling training data attribution that is ~500x faster than post hoc baselines when consuming equivalent memory. Calibrating linear prototype controllers can improve downstream accuracy by roughly 3 points while tracing those corrections back to training neighborhoods, and targeted prototype suppression can remove model behaviors without finetuning or measurable loss in generation quality.

Dan Ley, Giang Nguyen, Himabindu Lakkaraju +1

1Harvard University · 2Guide Labs Inc. *Work initiated during an internship at Guide Labs. · 2Guide Labs Inc.

LLM Interpretability Language Modeling Language Model Editing Training Data Attribution LLM Training

Jul 1, 2026cs.CL

A Mechanistic View of Authority Hierarchy in LLM Sycophancy

Authority bias poses a critical safety concern in language models: models systematically prioritize social cues from authority figures over factual consistency, swaying their answers based on source credibility rather than evidence. We mechanistically investigate this phenomenon using a controlled medical QA setting, where hints suggesting incorrect answers are attributed to personas of varying expertise. Across Llama-3.1-8B, Qwen3-8B, and Gemma-2-9B, we find that models respond in a graded manner proportional to perceived authority, a hierarchy that is never explicitly prompted but emerges from training. Logit lens analysis and linear/non-linear probing localize this effect to a critical late layer where correct answer representations are actively erased, an erasure that scales with authority level, resists mean vector intervention, and is only partially reversible through chain-of-thought reasoning. Our findings suggest that authority-induced sycophancy is not a surface-level output bias but mechanistic knowledge erasure, a precise, layer-localized overwriting of correct internal representations by high-status authority signals.

Emil Joswin, Srujananjali Medicherla, Priyanka Mary Mammen

Independent Research · University of Massachusetts Amherst

LLM Sycophancy LLM Interpretability

Jul 1, 2026q-bio.NC

NeuroCogMap Reveals Cognitive Organization of Large Language Models

Understanding how complex cognitive functions are organized within artificial systems is central to interpreting large language models (LLMs) and relating them to biological cognition. Yet although LLMs exhibit broad cognitive-like behaviours, it remains unclear whether their internal representations form reproducible functional systems that explain behaviour, failure and links to human cognition. Here we present NeuroCogMap, a cognitive neuroscience-inspired framework that organizes internal features of LLMs into functional parcels and links them to interpretable functions, cognitive capabilities and a cognitive hierarchy. These parcels form a stable and semantically coherent organization that is partly conserved across models and functionally linked to model outputs. Within this organization, major LLM failures, including hallucination, bias, refusal failure and sycophancy, correspond to distinct disruptions in representational and behavioural-control systems, yielding internal signatures for mechanism-guided detection and targeted intervention. Beyond model behaviour, NeuroCogMap improves prediction of human cortical responses during naturalistic language comprehension, with the strongest correspondence in higher-order association cortex. At the cognitive level, its internal signatures expose latent strategies that guide refinements of classical models of human decision-making. Together, these findings establish NeuroCogMap as a system-level framework for mapping functional organization in artificial systems and for relating this organization to human cortical function and cognitive behaviour.

Zhongxiang Sun, Haolang Lu, Qiang Ma +11

Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China · Beijing University of Posts and Telecommunications, Beijing, China · The University of Hong Kong, Hong Kong, China +5

Cognitive Modeling LLM Interpretability Representation Geometry in Language Models