{"total":16,"items":[{"citing_arxiv_id":"2606.20761","ref_index":5,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Integrating Large Language Model Agents with Digital Twins for Industrial Autonomous Systems","primary_cat":"cs.SE","submitted_at":"2026-06-18T09:48:44+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"A TPSR-based framework with four LLM roles integrates language model reasoning into industrial automation via digital twins, achieving high task executability in case studies.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.29358","ref_index":31,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet","primary_cat":"cs.AI","submitted_at":"2026-05-28T04:57:47+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Sparse autoencoders scaled to 34 million features on Claude 3 Sonnet yield interpretable, steerable representations of concrete and abstract concepts that generalize across languages and modalities.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.27970","ref_index":6,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Geometry of Human Perceptual Domains Emerges Transiently in LLM Representations","primary_cat":"cs.AI","submitted_at":"2026-05-27T05:04:29+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"Perceptual geometry for color, pitch, emotion and taste emerges transiently in intermediate layers of transformer LLMs despite purely textual training.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.07568","ref_index":88,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"A Systematic Study of Behavioral Cloning for Scientific Data Annotation","primary_cat":"cs.HC","submitted_at":"2026-05-26T02:19:47+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Introduces 9 synthetic annotation tasks and benchmarks for behavioral cloning, finding hierarchical skill learning, scaling benefits, effective multi-task pretraining, and shared internal representations of task phases and mistakes.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.19343","ref_index":4,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"What Makes a Representation Good for Single-Cell Perturbation Prediction?","primary_cat":"cs.LG","submitted_at":"2026-05-19T04:30:11+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"PerturbedVAE disentangles perturbation-specific signals from invariant gene expression structure to recover causal representations and improve out-of-distribution prediction in single-cell perturbation modeling.","context_count":1,"top_context_role":"background","top_context_polarity":"unclear","context_text":"plicit reference for background information. Specifically, for each perturbed sample(x,u), we sample a control ex- pression profilex (u0) from the unperturbed condition and encourage theirperturbation-invariantrepresentations to agree. Concretely, we align the invariant latents inferred from the two samples by minimizing Lcontrast(x,x (u0)) =∥z ι −z (u0) ι ∥2 2,(4) wherez ι ∼q θ(zι |x)andz (u0) ι ∼q θ(zι |x (u0)). Intuitively, by enforcing consistency ofz ι across per- turbed and unperturbed samples, the alignment enforces dominant, perturbation-invariant variation to be explained throughz ι. Once this dominant background variation is accounted for inz ι, it no longer needs to be explained by other latent variables during reconstruction."},{"citing_arxiv_id":"2605.19172","ref_index":10,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Bridge: Retrieval-Augmented Spatiotemporal Modeling for Urban Delivery Demand","primary_cat":"cs.LG","submitted_at":"2026-05-18T22:55:19+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Bridge augments a graph neural network backbone with time-aware retrieval from a memory of region-time windows to improve cold-start and cross-city urban delivery demand forecasting.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.12809","ref_index":259,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Correcting Influence: Unboxing LLM Outputs with Orthogonal Latent Spaces","primary_cat":"cs.LG","submitted_at":"2026-05-12T23:01:29+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"A latent mediation framework with sparse autoencoders enables non-additive token-level influence attribution in LLMs by learning orthogonal features and back-propagating attributions.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.01609","ref_index":48,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Concepts Whisper While Syntax Shouts: Spectral Anti-Concentration and the Dual Geometry of Transformer Representations","primary_cat":"cs.LG","submitted_at":"2026-05-02T21:20:55+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Transformer activations show spectral anti-concentration for concepts in the tail while syntax prefers high-variance directions, forming a dual geometry.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.25905","ref_index":15,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"A paradox of AI fluency","primary_cat":"cs.CL","submitted_at":"2026-04-28T17:51:13+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Fluent AI users adopt an active, iterative collaboration mode that produces more visible failures but better recovery and success on hard tasks, whereas novices experience more invisible failures from passive use.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"conversationalists (Grice, 1975) with a familiar sense of how common ground is established and negotiated (Clark et al., 1983). Experts quickly learn that this is not true; AIs' fluency tends to mask their fragmentary understanding of the context (Fried et al., 2023; Shao et al., 2025) and of the world (Vafa et al. 2024; 2025; Mancoridis et al. 2025; for alternative perspectives, see Gurnee & Tegmark 2024; Li et al. 2024; Tsvilodub et al. 2026). The specific augmentative behaviors that the current paper identifies in high-fluency users can be seen as indirect efforts to manage the unusual interactional nature of AI. The future of workDiscussions of AI fluency invite questions about how AI is likely to affect employment for workers in different industries and at different skill levels (Eloundou"},{"citing_arxiv_id":"2604.19052","ref_index":42,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Cell-Based Representation of Relational Binding in Language Models","primary_cat":"cs.CL","submitted_at":"2026-04-21T03:58:47+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Large language models encode relational bindings via a cell-based representation: a low-dimensional linear subspace in which each cell corresponds to an entity-relation index pair and attributes are retrieved from the matching cell.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.17105","ref_index":20,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"How Tokenization Limits Phonological Knowledge Representation in Language Models and How to Improve Them","primary_cat":"cs.CL","submitted_at":"2026-04-18T18:40:56+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"Subword tokenization impairs phonological knowledge encoding in LMs, but an IPA-based fine-tuning method restores it with minimal impact on other capabilities.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.00847","ref_index":8,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"H-Probes: Extracting Hierarchical Structures From Latent Representations of Language Models","primary_cat":"cs.CL","submitted_at":"2026-04-15T00:59:17+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"H-probes locate low-dimensional subspaces encoding hierarchy in LLM activations for synthetic tree tasks, show causal importance and generalization, and detect weaker signals in mathematical reasoning traces.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.06377","ref_index":22,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"The Master Key Hypothesis: Unlocking Cross-Model Capability Transfer via Linear Subspace Alignment","primary_cat":"cs.LG","submitted_at":"2026-04-07T19:02:10+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"The Master Key Hypothesis states that capabilities are low-dimensional directions transferable across models through linear subspace alignment, with UNLOCK demonstrating gains such as 12.1% accuracy improvement on MATH when transferring CoT from 14B to 7B models.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"Arora, S. Basart, E. Tang, D. Song, and J. Steinhardt. Measuringmathematicalproblemsolvingwiththemathdataset.arXivpreprintarXiv:2103.03874, 2021. URLhttps://arxiv.org/abs/2103.03874. [21] Y.Hong,D.Zhou,M.Cao,L.Yu,andZ.Jin. Thereasoning-memorizationinterplayinlanguage models is mediated by a single direction, 2025. URLhttps://arxiv.org/abs/2503.23084. [22] S.-C. Huang, P.-Z. Li, Y.-C. Hsu, K.-M. Chen, Y. T. Lin, S.-K. Hsiao, R. T.-H. Tsai, and H. yi Lee. Chat vector: A simple approach to equip llms with instruction following and model alignment in new languages, 2024. URLhttps://arxiv.org/abs/2310.04799. [23] Y. Huang, C. Huang, D. Feng, W. Lei, and J. Lv. Cross-model transferability among large language models on the platonic representations of concepts, 2025."},{"citing_arxiv_id":"2602.20338","ref_index":7,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Emergent Manifold Separability during Reasoning in Large Language Models","primary_cat":"cs.LG","submitted_at":"2026-02-23T20:36:17+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Reasoning in LLMs produces a transient geometric pulse in which concept manifolds untangle into linearly separable subspaces immediately before computation and compress afterward.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2406.04952","ref_index":13,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Quantifying Geospatial in the Common Crawl Corpus","primary_cat":"cs.CL","submitted_at":"2024-06-07T14:16:37+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Analysis estimates 18.7% of Common Crawl documents contain geospatial information like coordinates and addresses, with little difference by language.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2311.03658","ref_index":7,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"The Linear Representation Hypothesis and the Geometry of Large Language Models","primary_cat":"cs.CL","submitted_at":"2023-11-07T01:59:11+00:00","verdict":"CONDITIONAL","verdict_confidence":"MODERATE","novelty_score":8.0,"formal_verification":"none","one_line_summary":"Linear representations of high-level concepts in LLMs are formalized via counterfactuals in input and output spaces, unified under a causal inner product that enables consistent probing and steering.","context_count":1,"top_context_role":"background","top_context_polarity":"unclear","context_text":"counterfactual pairs, ⟨¯γW,(−i), γ(yi(1)) − γ(yi(0))⟩C, are shown in red. For reference, we also project the differences between 100K randomly sampled word pairs onto the estimated concept direction, as shown in blue. See Table 2 for details about each concept W (the title of each plot). 3 6 9 12 15 18 21 24 27 verb 3pSg (1) verb Ving (2) verb Ved (3) Ving 3pSg (4) Ving Ved (5) 3pSg Ved (6) verb V + able (7) verb V + er (8) verb V + tion (9) verb V + ment (10) adj un + adj (11) adj adj + ly (12) small big (13) thing color (14) thing part (15) country capital (16) pronoun possessive (17) male female (18) lower upper (19) noun plural (20) adj comparative (21) adj superlative (22) frequent infrequent (23) English French (24) French German (25) French Spanish (26)"}],"limit":50,"offset":0}