Language Model Probing

Perturbation probing generates task-specific causal hypotheses for FFN neurons in large language models using two forward passes per prompt and no backpropagation, followed by a one-time intervention sweep of about 150 passes amortized across all identified neurons. Across eight behavioral circuits, 13 models, and four architecture families, we identify two circuit structures that organize LLM behavior. Opposition circuits appear when RLHF suppresses a pre-training tendency. In safety refusal, about 50 neurons, or 0.014 percent of all neurons, control the refusal template; ablating them changes 80 percent of response formats on 520 AdvBench prompts while producing near-zero harmful compliance, 3 of 520 cases, all with disclaimers. Routing circuits appear for pre-training behaviors distributed through attention. For language selection, residual-stream direction injection switches English to Chinese output on 99.1 percent of 580 benchmark prompts in the 3 of 19 tested models that satisfy three observed conditions: bilingual training, FFN-to-skip signal ratio between 0.3 and 1.1, and linear representability. The same intervention fails on the other 16 models and on math, code, and factual circuits, defining the limits of directional steering. The FFN-to-skip signal ratio, computed from the same two forward passes, distinguishes the two structures and predicts the appropriate intervention. Circuit topology varies by architecture, from Qwen's concentrated FFN bottleneck to Gemma's normalization-shielded circuit. In Qwen3.5-2B, ablating 20 neurons eliminates multi-turn sycophantic capitulation, while amplifying 10 related neurons improves factual correction from 52 percent to 88 percent on 200 TruthfulQA prompts. These results show that perturbation probing offers mechanistic insight into RLHF-organized behavior and a practical toolkit for precision template-layer editing.

Do large language models (LLMs) truly acquire embodied cognition and cultural conventions from text? We introduce demonstratives, fundamental spatial expressions like "this/that" in English and "zhè/nà" in Chinese, as a novel probe for grounded knowledge. Using 6,400 responses from 320 native speakers, we establish a human baseline: English speakers reliably distinguish proximal-distal referents but struggle with perspective-taking, while Chinese speakers switch perspectives fluently but tolerate distal ambiguity. In contrast, five state-of-the-art LLMs fail to inherently understand the proximal-distal contrast and show no cultural differences, defaulting to English-centric reasoning. Our study contributes (i) a new task, based on demonstratives, as a new lens for evaluating embodied cognition and cultural conventions; (ii) empirical evidence of cross-cultural asymmetries in human interpretation; (iii) a new perspective on the egocentric-sociocentric debate, showing both orientations coexist but vary across languages; and (iv) a call to address individual variation in future model design.

Auditing the fine-tunes of open-weight generative models for harmful specialization has become a new governance challenge for model hosting platforms. The standard toolkit, generative evaluation via curated prompts or red-teaming, does not scale to platform-level auditing and breaks down entirely for domains like CSAM where generation is legally constrained. This motivates the Evaluation without Generation problem: assessing model capabilities without producing outputs. We argue that in such settings, capability must be inferred from the model's state, either its parameters or internal representations, rather than its outputs. We introduce Gaussian probing, a method that characterizes how LoRA adaptors perturb a model's internal representations by measuring responses to Gaussian latent ensembles. Unlike raw-weight baselines, Gaussian probing reliably distinguishes benign from harmful specialization without sampling outputs. We demonstrate effectiveness in high-risk domains, including detecting models specialized for child sexual abuse material (CSAM), where output-based evaluation is legally and ethically constrained. Our results show that Gaussian probing provides a scalable non-generative alternative for evaluating high-risk generative systems and remains robust to weight rescaling, a representative adversarial manipulation.

Closed-source frontier labs do not disclose parameter counts. Storing F facts requires at least F/(bits per parameter) weights, so factual recall lower-bounds parameter count--an intrinsic, serving-independent signal, though (as we show) a coarse one. We introduce Incompressible Knowledge Probes (IKPs), a benchmark of 1,400 factual questions spanning 7 tiers of obscurity, designed to isolate knowledge that cannot be derived by reasoning or compressed by architectural improvements. We score with no hallucination penalty (lambda = 0: IKP accuracy is simply the fraction of probed facts answered correctly), which removes both the penalty hyperparameter and the per-tier flooring choice; a full lambda x flooring ablation shows the calibration is robust across scoring choices while individual estimates are not, motivating the no-penalty default. We calibrate a log-linear mapping from IKP accuracy to parameter count on 93 open-weight models (135M-1,600B) spanning 19 vendors, achieving R^2 = 0.910; leave-one-out cross-validation confirms generalization (median fold error 1.48x, 72% within 2x, 86% within 3x). The instrument is deliberately coarse--its 90% prediction interval spans ~3x in either direction, wider than inference economics--so IKP recovers order-of-magnitude effective capacity and relative rankings, not precise parameter counts. For Mixture-of-Experts models, total parameters predict knowledge (R^2 = 0.67) better than active parameters (R^2 = 0.41). We evaluate 201 models from 27 vendors on a curated probe set (1,311 of 1,400 probes surviving name-collision and label-ambiguity filters) and report effective knowledge capacity for all major proprietary frontier models as prediction bands rather than point estimates; for heavily safety-tuned models these are lower bounds, since refusal policy can suppress tens of percentage points of otherwise-answerable capacity.

Large Language Models (LLMs) reveal inherent and distinctive personas through dialogue. However, most existing persona discovery approaches rely on surface-level lexical or stylistic cues, treating dialogue as a flat sequence of tokens and failing to capture the deeper discourse-level structures that sustain persona consistency. To address this limitation, we propose a novel analytical framework that interprets LLM dialogue through bridging inference -- implicit conceptual relations that connect utterances via shared world knowledge and discourse coherence. By modeling these relations as structured knowledge graphs, our approach captures latent semantic links that govern how LLMs organize meaning across turns, enabling persona discovery at the level of discourse coherence rather than surface realizations. Experimental results across multiple reasoning backbones and target LLMs, ranging from small-scale models to 80B-parameter systems, demonstrate that bridging-inference graphs yield significantly stronger semantic coherence and more stable persona identification than frequency or style-based baselines. These results show that persona traits are consistently encoded in the structural organization of discourse rather than isolated lexical patterns. This work presents a systematic framework for probing, extracting, and visualizing latent LLM personas through the lens of Cognitive Discourse Theory, bridging computational linguistics, cognitive semantics, and persona reasoning in large language models. Codes are available at https://github.com/JiSoo-Yang/Persona_Bridging.git

Understanding what kinds of factual knowledge large language models (LLMs) memorize is essential for evaluating their reliability and limitations. Entity-based QA is a common framework for analyzing non-verbatim memorization, but typical evaluations query each entity using a single canonical surface form, making it difficult to disentangle fact memorization from access through a particular name. We introduce RedirectQA, an entity-based QA dataset that uses Wikipedia redirect information to associate Wikidata factual triples with categorized surface forms for each entity, including alternative names, abbreviations, spelling variants, and common erroneous forms. Across 13 LLMs, we examine surface-conditioned factual memorization and find that prediction outcomes often change when only the entity surface form changes. This inconsistency is category-dependent: models are more robust to minor orthographic variations than to larger lexical variations such as aliases and abbreviations. Frequency analyses further suggest that both entity- and surface-level frequencies are associated with accuracy, and that entity frequency often contributes beyond surface frequency. Overall, factual memorization appears neither purely surface-specific nor fully surface-invariant, highlighting the importance of surface-form diversity in evaluating non-verbatim memorization.

Large language models increasingly shape the information people consume: they are embedded in search, consulted for professional advice, deployed as agents, and used as a first stop for questions about policy, ethics, health, and politics. When such a model silently holds a position on a contested topic, that position propagates at scale into users' decisions. Eliciting a model's positions is harder than it first appears: contemporary assistants answer direct opinion questions with evasive disclaimers, and the same model may concede the opposite position once the user starts arguing one side. We propose a method, released as the open-source llm-bias-bench, for discovering the opinions an LLM actually holds on contested topics under conditions that resemble real multi-turn interaction. The method pairs two complementary free-form probes. Direct probing asks for the model's opinion across five turns of escalating pressure from a simulated user. Indirect probing never asks for an opinion and engages the model in argumentative debate, letting bias leak through how it concedes, resists, or counter-argues. Three user personas (neutral, agree, disagree) collapse into a nine-way behavioral classification that separates persona-independent positions from persona-dependent sycophancy, and an auditable LLM judge produces verdicts with textual evidence. The first instantiation ships 38 topics in Brazilian Portuguese across values, scientific consensus, philosophy, and economic policy. Applied to 13 assistants, the method surfaces findings of practical interest: argumentative debate triggers sycophancy 2-3x more than direct questioning (median 50% to 79%); models that look opinionated under direct questioning often collapse into mirroring under sustained arguments; and attacker capability matters mainly when an existing opinion must be dislodged, not when the assistant starts neutral.

Language models must bind entities to the attributes they possess and maintain several such binding relationships within a context. We study how multiple entities are represented across token positions and whether single tokens can carry bindings for more than one entity. We introduce a multi-slot probing approach that disentangles a single token's residual stream activation to recover information about both the currently described entity and the immediately preceding one. These two kinds of information are encoded in separate and largely orthogonal "current-entity" and "prior-entity" slots. We analyze the functional roles of these slots and find that they serve different purposes. In tandem with the current-entity slot, the prior-entity slot supports relational inferences, such as entity-level induction ("who came after Alice in the story?") and conflict detection between adjacent entities. However, only the current-entity slot is used for explicit factual retrieval questions ("Is anyone in the story tall?" "What is the tall entity's name?") despite these answers being linearly decodable from the prior-entity slot too. Consistent with this limitation, open-weight models perform near chance accuracy at processing syntax that forces two subject-verb-object bindings on a single token (e.g., "Alice prepares and Bob consumes food.") Interestingly, recent frontier models can parse this properly, suggesting they may have developed more sophisticated binding strategies. Overall, our results expose a gap between information that is available in activations and information the model actually uses, and suggest that the current/prior-entity slot structure is a natural substrate for behaviors that require holding two perspectives at once, such as sycophancy and deception.

We study how large language models recall relational knowledge during text generation, with a focus on identifying latent representations suitable for relation classification via linear probes. Prior work shows how attention heads and MLPs interact to resolve subject, predicate, and object, but it remains unclear which representations support faithful linear relation classification and why some relation types are easier to capture linearly than others. We systematically evaluate different latent representations derived from attention head and MLP contributions, showing that per-head attention contributions to the residual stream are comparatively strong features for linear relation classification. Feature attribution analyses of the trained probes, as well as characteristics of the different relation types, reveal clear correlations between probe accuracy and relation specificity, entity connectedness, and how distributed the signal on which the probe relies is across attention heads. Finally, we show how token-level feature attribution of probe predictions can be used to reveal probe behavior in further detail.

Aligned language models refuse harmful instructions, but the representations through which they recognise such instructions are less well characterised than the behaviours they produce. Harmful intent is linearly separable from residual-stream activations across 12 models spanning four architectural families (Qwen2.5, Qwen3.5, Llama-3.2, Gemma-3) and three alignment variants (base, instruction-tuned, abliterated), with parameter scales from 0.5B to 1.3B and a within-family scale extension to 9B on Qwen3.5. A direction fitted from 100 labelled examples per class via Soft-AUC optimisation reaches mean effective AUROC 0.982 and TPR@1%FPR 0.797, generalises to three held-out harm benchmarks and a hard-benign control, and matches its instruction-tuned counterpart within $\pm 0.003$ AUROC in abliterated variants from which the refusal mechanism has been removed. The supervised strategies all exceed AUROC 0.96, but their TPR@1%FPR varies by more than ten times the AUROC gap; a deployed 9B safety classifier shows the same pattern at AUROC 0.94 and TPR 0.30, motivating low-FPR reporting as a default in safety-adjacent detection evaluation. Geometric measurements refine the picture. The recovered direction is concentrated within each extraction protocol but protocol-dependent across them: two pooling choices applied to the same chat-templated activations at the same residual-stream layer (max-pool over content tokens versus last-token at the post-instruction position) recover harm directions $73^\circ$ apart, and projecting one out leaves detection under either max-pool extraction essentially intact. Probing identifies a protocol-specific direction rather than a unique computational feature.

Probing has shown that language model representations encode rich linguistic information, but it remains unclear whether they also capture cognitive signals about human processing. In this work, we probe language model representations for human reading times. Using regularized linear regression on two eye-tracking corpora spanning five languages (English, Greek, Hebrew, Russian, and Turkish), we compare the representations from every model layer against scalar predictors -- surprisal, information value, and logit-lens surprisal. We find that the representations from early layers outperform surprisal in predicting early-pass measures such as first fixation and gaze duration. The concentration of predictive power in the early layers suggests that human-like processing signatures are captured by low-level structural or lexical representations, pointing to a functional alignment between model depth and the temporal stages of human reading. In contrast, for late-pass measures such as total reading time, scalar surprisal remains superior, despite its being a much more compressed representation. We also observe performance gains when using both surprisal and early-layer representations. Overall, we find that the best-performing predictor varies strongly depending on the language and eye-tracking measure.

A recent study (Kuribayashi et al., 2025) has shown that human sentence processing behavior, typically measured on syntactically unchallenging constructions, can be effectively modeled using surprisal from early layers of large language models (LLMs). This raises the question of whether such advantages of internal layers extend to more syntactically challenging constructions, where surprisal has been reported to underestimate human cognitive effort. In this paper, we begin by exploring internal layers that better estimate human cognitive effort observed in syntactic ambiguity processing in English. Our experiments show that, in contrast to naturalistic reading, later layers better estimate such a cognitive effort, but still underestimate the human data. This dual alignment sheds light on different modes of sentence processing in humans and LMs: naturalistic reading employs a somewhat weak prediction akin to earlier layers of LMs, while syntactically challenging processing requires more fully-contextualized representations, better modeled by later layers of LMs. Motivated by these findings, we also explore several probability-update measures using shallow and deep layers of LMs, showing a complementary advantage to single-layer's surprisal in reading time modeling.

Static concreteness ratings are widely used in NLP, yet a word's concreteness can shift with context, especially in figurative language such as metaphor, where common concrete nouns can take abstract interpretations. While such shifts are evident from context, it remains unclear how LLMs understand concreteness internally. We conduct a layer-wise and geometric analysis of LLM hidden representations across four model families, examining how models distinguish literal vs figurative uses of the same noun and how concreteness is organized in representation space. We find that LLMs separate literal and figurative usage in early layers, and that mid-to-late layers compress concreteness into a one-dimensional direction that is consistent across models. Finally, we show that this geometric structure is practically useful: a single concreteness direction supports efficient figurative-language classification and enables training-free steering of generation toward more literal or more figurative rewrites.

Benchmarks are often used as a standard to understand LLM capabilities in different domains. However, aggregate benchmark scores provide limited insight into compositional skill gaps of LLMs and how to improve them. To make these weaknesses visible, we propose Scaffolded Task Design (STaD) framework. STaD generates controlled variations of benchmark tasks based on the concept of scaffolding, which introduces structured, incremental support in a step-by-step manner. Rather than inspecting failures individually, this approach enables systematic and scalable probing of model behavior by identifying the specific reasoning skill compositions they lack. Treating the LLM as a black box, our experiments on six models of varying sizes reveal multiple failure points in three reasoning benchmarks and highlight each model's unique and distinct skill gaps.

Large language models (LLMs) have achieved strong performance on natural language to SQL (NL2SQL) benchmarks, yet their reported accuracy may be inflated by contamination from benchmark queries or structurally similar patterns seen during training. We introduce SPENCE (Syntactic Probing and Evaluation of NL2SQL Contamination Effects), a controlled syntactic probing framework for detecting and quantifying such contamination. SPENCE systematically generates syntactic variants of test queries for four widely used NL2SQL datasets-Spider, SParC, CoSQL, and the newer BIRD benchmark. We use SPENCE to evaluate multiple high-capacity LLMs under execution-based scoring. For each model, we measure changes in execution accuracy across increasing levels of syntactic divergence and quantify rank sensitivity using Kendall's tau with bootstrap confidence intervals. By aligning these robustness trends with benchmark release dates, we observe a clear temporal gradient: older benchmarks such as Spider exhibit the strongest negative values and thus the highest likelihood of training leakage, whereas the more recent BIRD dataset shows minimal sensitivity and appears largely uncontaminated. Together, these findings highlight the importance of temporally contextualized, syntactic-probing evaluation for trustworthy NL2SQL benchmarking.

Clinical personality assessment screens response validity before interpreting substantive scales. LLM evaluation does not. We apply the validity scaling framework from the PAI and MMPI-3 to metacognitive probe data from 20 frontier models across 524 items. Six validity indices are operationalised: L (maintaining confidence on errors), K (betting on errors), F (withdrawing consensus-endorsed items), Fp (withdrawing correct answers), RBS (inverted monitoring), and TRIN (fixed responding). A tiered classification system identifies four models as construct-level invalid and two as elevated. Valid-profile models produce item-sensitive confidence (mean r = .18, 14 of 16 significant). Invalid-profile models do not (mean r = -.20, d = 2.17, p = .001). Chain-of-thought training produces two opposite response distortions. Two latent dimensions account for 94.6% of index variance. Companion papers extract a portable screening protocol (Cacioli, 2026e) and validate it against selective prediction (Cacioli, 2026f). All data and code: https://github.com/synthiumjp/validity-scaling-llm

When model developers or users fine-tune an LLM, this can induce behaviors that are unexpected, deliberately harmful, or hard to detect. It would be far easier to audit LLMs if they could simply describe their behaviors in natural language. Here, we study a scalable approach to rapidly identify learned behaviors of many LLMs derived from a shared base LLM. Given a model $M$, our method works by finetuning models $M_i$ from $M$ with implanted behaviors $b_i$; the $(M_i, b_i)$ pairs serve as labeled training data. We then train an introspection adapter (IA): a single LoRA adapter jointly trained across the finetunes $M_i$ to cause them to verbalize their implanted behaviors. We find that this IA induces self-description of learned behaviors even in finetunes of $M$ that were trained in very different ways from the $M_i$. For example, IAs generalize to AuditBench, achieving state-of-the-art at identifying explicitly hidden concerning behaviors. IAs can also be used to detect encrypted finetuning API attacks. They scale favorably with model size and training data diversity. Overall, our results suggest that IAs are a scalable, effective, and practically useful approach to auditing fine-tuned LLMs.

Large Language Models (LLMs) are increasingly exposed to adaptive jailbreaking, particularly in high-stakes Chemical, Biological, Radiological, and Nuclear (CBRN) domains. Although streaming probes enable real-time monitoring, they still make systematic errors. We identify a core issue: existing methods often rely on a few high-scoring tokens, leading to false alarms when sensitive CBRN terms appear in benign contexts. To address this, we introduce a streaming probing objective that requires multiple evidence tokens to consistently support a prediction, rather than relying on isolated spikes. This encourages more robust detection based on aggregated signals instead of single-token cues. At a fixed 1% false-positive rate, our method improves the true-positive rate by 35.55% relative to strong streaming baselines. We further observe substantial gains in AUROC, even when starting from near-saturated baseline performance (AUROC = 97.40%). We also show that probing Attention or MLP activations consistently outperforms residual-stream features. Finally, even when adversarial fine-tuning enables novel character-level ciphers, harmful intent remains detectable: probes developed for the base LLMs can be applied ``plug-and-play'' to these obfuscated attacks, achieving an AUROC of over 98.85%.

Linear probes can detect when language models produce outputs they "know" are wrong, a capability relevant to both deception and reward hacking. However, single-layer probes are fragile: the best layer varies across models and tasks, and probes fail entirely on some deception types. We show that combining probes from multiple layers into an ensemble recovers strong performance even where single-layer probes fail, improving AUROC by +29% on Insider Trading and +78% on Harm-Pressure Knowledge. Across 12 models (0.5B--176B parameters), we find probe accuracy improves with scale: ~5% AUROC per 10x parameters (R=0.81). Geometrically, deception directions rotate gradually across layers rather than appearing at one location, explaining both why single-layer probes are brittle and why multi-layer ensembles succeed.

Representing and navigating hierarchy is a fundamental primitive of reasoning. Large language models have demonstrated proficiency in a wide variety of tasks requiring hierarchical reasoning, but there exists limited analysis on how the models geometrically represent the necessary latent constructions for such thinking. To this end, we develop H-probes, a collection of linear probes that extract hierarchical structure, specifically depth and pairwise distance, from latent representations. In synthetic tree traversal tasks, the H-probes robustly find the subspaces containing hierarchical structure necessary to complete the tasks; furthermore, in comprehensive ablation experiments, we show that these hierarchy-containing subspaces are low-dimensional, causally important for high task performance, and generalize within- and out-of-domain. Furthermore, we find analogous, though weaker, hierarchical structure in real-world hierarchical contexts such as mathematical reasoning traces. These results demonstrate that models represent hierarchy not only at the level of syntax and concepts, but at deeper levels of abstraction -- including the reasoning process itself.

Narrative understanding requires multidimensional semantic structures. This study investigates whether BERT embeddings encode dimensions of fictional narrative semantics -- time, space, causality, and character. Using an LLM to accelerate annotation, we construct a token-level dataset labeled with these four narrative categories plus "others." A linear probe on BERT embeddings (94% accuracy) significantly outperforms a control probe on variance-matched random embeddings (47%), confirming that BERT encodes meaningful narrative information. With balanced class weighting, the probe achieves a macro-average recall of 0.83, with moderate success on rare categories such as causality (recall = 0.75) and space (recall = 0.66). However, confusion matrix analysis reveals "Boundary Leakage," where rare dimensions are systematically misclassified as "others." Clustering analysis shows that unsupervised clustering aligns near-randomly with predefined categories (ARI = 0.081), suggesting that narrative dimensions are encoded but not as discretely separable clusters. Future work includes a POS-only baseline to disentangle syntactic patterns from narrative encoding, expanded datasets, and layer-wise probing.

Large language models are increasingly used as writing tools and pedagogical resources in English for Academic Purposes, but it remains unclear whether they encode culturally differentiated representations when generating academic text. This study tests whether Gemma-3-4b-it encodes nationality-discriminative information in hidden states when generating research article introductions conditioned by British and Chinese academic personas. A corpus of 270 texts was generated from 45 prompt templates crossed with six persona conditions in a 2 x 3 design. Logistic regression probes were trained on hidden-state activations across all 35 layers, with shuffled-label baselines, a surface-text skyline classifier, cross-family tests, and sentence-level baselines used as controls. Probe-selected token positions were annotated for structural, lexical, and stance features using the Stanza NLP pipeline. The nationality probe reached 0.968 cross-validated accuracy at Layer 18, with perfect held-out classification. Nationality encoding followed a non-monotonic trajectory across layers, with structural effects strongest in the middle to upper network and lexical-domain effects peaking earlier. At high-signal token positions, British-associated patterns showed more postmodification, hedging, boosting, passive voice, and evaluative or process-oriented vocabulary, while Chinese-associated patterns showed more premodification, nominal predicates, and sociocultural or internationalisation vocabulary. However, sentence-level analysis found no significant nationality differences in the full generated surface text. The findings extend probing methodology to a sociolinguistic attribute and have practical implications for EAP and language pedagogy.

Do multilingual embedding models encode a language-general representation of proficiency? We investigate this by training linear and non-linear probes on hidden-state activations from seven embedding models (0.3-8B) to predict CEFR proficiency levels from learner texts across nine corpora and seven languages. We compare five probing architectures against a baseline trained on surface-level text features. Under in-distribution evaluation, probes achieve strong performance (Quadratic Weighted Kappa $\approx0.7$), substantially outperforming the surface baseline, with middle layers consistently yielding the best predictions. However, in cross-corpus evaluation performance collapses across all probe types and model sizes. Residual analysis reveals that out-of-distribution probes converge towards predicting uniformly distributed labels, indicating that the learned mappings capture corpus-specific distributional properties (topic, language, task type, rating methodology) rather than an abstract, transferable proficiency dimension. These results suggest that current multilingual embeddings do not straightforwardly encode language-general proficiency, with implications for representation-based approaches to proficiency-adaptive language technology.

Large language models (LLMs) frequently encode factual and reasoning knowledge in their internal representations that is not faithfully reflected in their surface-level outputs -- a phenomenon known as latent knowledge. Existing approaches to eliciting latent knowledge, such as Contrastive Consistency Search (CCS), rely on contrastive activation patterns and struggle with complex multi-step reasoning tasks, while mechanistic interpretability tools have primarily been used to understand model behavior rather than to extract hidden knowledge. We present MechELK, a unified three-stage framework that bridges mechanistic interpretability and latent knowledge elicitation. MechELK operates through: (1) Locate -- using Sparse Autoencoder (SAE) feature analysis and activation patching to identify knowledge-bearing representations; (2) Verify -- employing causal probing to distinguish genuine latent knowledge from spurious correlations; and (3) Elicit -- applying representation engineering to surface hidden knowledge without modifying model weights. Evaluated on TruthfulQA, a curated Deceptive Alignment benchmark, and the Quirky LM dataset, MechELK achieves an average elicitation accuracy of 84.7%, outperforming CCS by 6.2% and direct linear probing by 9.1%. Crucially, MechELK successfully identifies latent knowledge in 78.3% of cases where the model's surface output is incorrect or evasive, demonstrating its utility for AI safety applications including deceptive alignment detection.

Existing hallucination detection methods for large language models (LLMs) rely on external verification at inference time, requiring gold answers, retrieval systems, or auxiliary judge models. We ask whether this external supervision can instead be distilled into the model's own representations during training, enabling hallucination detection from internal activations alone at inference time. We introduce a weak supervision framework that combines three complementary grounding signals: substring matching, sentence embedding similarity, and an LLM as a judge verdict to label generated responses as grounded or hallucinated without human annotation. Using this framework, we construct a 15000-sample dataset from SQuAD v2 (10500 train/development samples and a separate 5000-sample test set), where each example pairs a LLaMA-2-7B generated answer with its full per-layer hidden states and structured hallucination labels. We then train five probing classifiers: ProbeMLP (M0), LayerWiseMLP (M1), CrossLayerTransformer (M2), HierarchicalTransformer (M3), and CrossLayerAttentionTransformerV2 (M4), directly on these hidden states, treating external grounding signals as training-time supervision only. Our central hypothesis is that hallucination detection signals can be distilled into transformer representations, enabling internal detection without any external verification at inference time. Results support this hypothesis. Transformer-based probes achieve the strongest discrimination, with M2 performing best on 5-fold average AUC/F1, and M3 performing best on both single-fold validation and held-out test evaluation. We also benchmark inference efficiency: probe latency ranges from 0.15 to 5.62 ms (batched) and 1.55 to 6.66 ms (single sample), while end-to-end generation plus probe throughput remains approximately 0.231 queries per second, indicating negligible practical overhead.

As the influence of LLMs expands, it is imperative to gain insight into their decisions. One way to do that is to develop probes that detect the presence or absence of a broad set of high-level abstract concepts within the embeddings computed in an LLM - which is what we might say a model is ``thinking" about. Such probes should be low-cost and easily applicable to any LLM, so that monitoring for many concepts is possible during normal operation. In this paper, we take the first steps towards developing the capability of creating many such probes by defining and executing examples of the key tasks needed: first, the careful delineation of a high-level abstract concept through the creation of a dataset with the concept both present and then absent. Then, the training and testing of a set of linear probes to detect the concept on any layer of an LLM, including an exploration of the complexity of the probe needed. Finally, we show that such probes can track concepts across larger contexts. This is done with four separate concepts and three different LLMs. When this process is scaled to many more concepts, it will create the ability to monitor new models.

Analogical reasoning is a core cognitive faculty essential for narrative understanding. While LLMs perform well when surface and structural cues align, they struggle in cases where an analogy is not apparent on the surface but requires latent information, suggesting limitations in abstraction and generalisation. In this paper we compare a model's probed representations with its prompted performance at detecting narrative analogies, revealing an asymmetry: for rhetorical analogies, probing significantly outperforms prompting in open-source models, while for narrative analogies, they achieve a similar (low) performance. This suggests that the relationship between internal representations and prompted behavior is task-dependent and may reflect limitations in how prompting accesses available information.

Interpretability research has highlighted the importance of evaluating Pretrained Language Models (PLMs) and in particular contextual embeddings against explicit linguistic theories to determine what linguistic information they encode. This study focuses on the Italian NPN (noun-preposition-noun) constructional family, challenging some of the theoretical and methodological assumptions underlying previous experimental designs and extending this type of research to a lesser-investigated language. Contextual vector representations are extracted from BERT and used as input to layer-wise probing classifiers, systematically evaluating information encoded across the model's internal layers. The results shed light on the extent to which constructional form and meaning are reflected in contextual embeddings, contributing empirical evidence to the dialogue between constructionist theory and neural language modelling

Detecting whether a model's internal knowledge is sufficient to correctly answer a given question is a fundamental challenge in deploying responsible LLMs. In addition to verbalising the confidence by LLM self-report, more recent methods explore the model internals, such as the hidden states of the response tokens, to capture how much knowledge is activated. We argue that such activated knowledge may not align with what the query requires, e.g., capturing the stylistic and length-related features that are uninformative for answering the query. To fill the gap, we propose GRADE (GRAdient Dynamics for knowlEdge gap detection), which quantifies the knowledge gap via the cross-layer rank ratio of the gradient to that of the corresponding hidden state subspace. This is motivated by the property of gradients as estimators of the required knowledge updates for a given target. We validate GRADE on six benchmarks, demonstrating its effectiveness and robustness to input perturbations. In addition, we present a case study demonstrating how the gradient chain can generate interpretable explanations of knowledge gaps in long-form answers. Code is available at https://github.com/yjEugenia/llm-aware

Standard LLM benchmarks evaluate the assistant turn: the model generates a response to an input, a verifier scores correctness, and the analysis ends. This paradigm leaves unmeasured whether the LLM encodes any awareness of what follows the assistant response. We propose user-turn generation as a probe of this gap: given a conversation context of user query and assistant response, we let a model generate under the user role. If the model's weights encode interaction awareness, the generated user turn will be a grounded follow-up that reacts to the preceding context. Through experiments across $11$ open-weight LLMs (Qwen3.5, gpt-oss, GLM) and $5$ datasets (math reasoning, instruction following, conversation), we show that interaction awareness is decoupled from task accuracy. In particular, within the Qwen3.5 family, GSM8K accuracy scales from $41\%$ ($0.8$B) to $96.8\%$ ($397$B-A$17$B), yet genuine follow-up rates under deterministic generation remain near zero. In contrast, higher temperature sampling reveals interaction awareness is latent with follow up rates reaching $22\%$. Controlled perturbations validate that the proposed probe measures a real property of the model, and collaboration-oriented post-training on Qwen3.5-2B demonstrates an increase in follow-up rates. Our results show that user-turn generation captures a dimension of LLM behavior, interaction awareness, that is unexplored and invisible with current assistant-only benchmarks.