Latest 50 Papers - July 23, 2026

**Please check the [Github](https://github.com/NeoFii/Daily_ArXiv) page for a better reading experience and more papers.**

## LLM Reasoning
| **Title** | **Date** | **Comment** |
| --- | --- | --- |
| **[They'll Verify. They Just Won't Act. How Authority Framing and Laundered Code Turn a Trusted Agentic CI/CD Pipeline Into an Attack Surface](https://arxiv.org/abs/2607.19267v1)** | 2026-07-21 | <details><summary>9 pag...</summary>9 pages. Dataset and reproduction code: https://github.com/senthex-security/senthex-research</details> |
| **[Reasoning Error from Known Fact: Step-Level Self-Consistency Group Relative Policy Optimization for LLM](https://arxiv.org/abs/2607.18915v1)** | 2026-07-21 | |
| **[Tracing the Shadows: Automatic Tracking and Analysis of Crypto Money Laundering via Transaction Semantic Analysis](https://arxiv.org/abs/2607.18869v1)** | 2026-07-21 | |
| **[Not All Errors Are Created Equal: ASCoT Addresses Late-Stage Fragility in Efficient LLM Reasoning](https://arxiv.org/abs/2508.05282v6)** | 2026-07-21 | |
| **[LLM-Grounded Explainable AI for Supply Chain Risk Early Warning via Temporal Graph Attention Networks](https://arxiv.org/abs/2603.04818v3)** | 2026-07-21 | |
| **[QuArch: A Benchmark for Evaluating LLM Reasoning in Computer Architecture](https://arxiv.org/abs/2510.22087v2)** | 2026-07-21 | |
| **[Oracle Gap and Signal Fidelity: A Fixed-Pool Diagnostic for Test-Time Collaboration](https://arxiv.org/abs/2607.17531v1)** | 2026-07-20 | <details><summary>13 pa...</summary>13 pages, 2 figures, 8 tables. Code: https://github.com/AmGarfield/OracleGap</details> |
| **[LLMs as a Jury: Cross-Model Consensus Can Outperform Process Reward Models for LLM Reasoning](https://arxiv.org/abs/2607.10139v2)** | 2026-07-20 | |
| **[AIGB-R1: Self-Evolving Generative Auto-Bidding via Hierarchical Planner-Executor Optimization](https://arxiv.org/abs/2607.17281v1)** | 2026-07-19 | |
| **[Debate-on-Graph: Reliable and Adaptive Reasoning of Large Language Model on Uncertain Knowledge Graph](https://arxiv.org/abs/2607.17266v1)** | 2026-07-19 | <details><summary>18 pa...</summary>18 pages, Accepted by ECML-PKDD 2026</details> |
| **[Constrained Path Reasoning: Measuring When Committed Stages Earn Their Cost](https://arxiv.org/abs/2607.17240v1)** | 2026-07-19 | |
| **[Prismatic Synthesis: Gradient-based Data Diversification Boosts Generalization in LLM Reasoning](https://arxiv.org/abs/2505.20161v2)** | 2026-07-19 | |
| **[Logic-Guided Socially-aware Robot Navigation World Model](https://arxiv.org/abs/2510.23509v3)** | 2026-07-18 | <details><summary>The a...</summary>The authors have decided to withdraw this manuscript due to concerns regarding its current scope, framing, and presentation. Please do not cite this version</details> |
| **[NOWJ@COLIEE 2026: Adaptive Pipelines for Legal Retrieval and Reasoning](https://arxiv.org/abs/2607.16603v1)** | 2026-07-18 | <details><summary>Prese...</summary>Presented at COLIEE 2026</details> |
| **[Video Streaming Thinking: VideoLLMs Can Watch and Think Simultaneously](https://arxiv.org/abs/2603.12262v2)** | 2026-07-17 | <details><summary>Accep...</summary>Accepted by ECCV 2026, project page https://1ranguan.github.io/VST/</details> |
| **[REST: Receding Horizon Explorative Steiner Tree for Zero-Shot Object-Goal Navigation](https://arxiv.org/abs/2603.18624v2)** | 2026-07-16 | Accepted to IROS'26 |
| **[MARS: Multi-hop Adaptive Retrieval and SPARQL Generation for KGQA](https://arxiv.org/abs/2607.14561v1)** | 2026-07-16 | <details><summary>EKAW ...</summary>EKAW 2026 (https://ekaw2026.di.unito.it/accepted-posters-and-demos)</details> |
| **[ProfMalPlus: Agent-Coordinated Detection of Malicious NPM Packages via Static-Dynamic Analysis Synergy](https://arxiv.org/abs/2607.13965v1)** | 2026-07-15 | |
| **[Policy of Thoughts: Scaling Test-Time Training for LLM Reasoning via Online Policy Evolution](https://arxiv.org/abs/2601.20379v2)** | 2026-07-15 | <details><summary>Under...</summary>Under Review, preprint</details> |
| **[Assert, don't describe: Linguistic features that shift LLM reasoning about animal welfare](https://arxiv.org/abs/2606.26104v3)** | 2026-07-12 | |
| **[Cognitive Episodes in LLM Reasoning Traces Enable Interpretable Human Item Difficulty Prediction](https://arxiv.org/abs/2606.28186v2)** | 2026-07-11 | <details><summary>32 pa...</summary>32 pages, 8 figures, 10 tables</details> |
| **[Can AI Reason Like an Urban Planner? Benchmarking Large Language Models Against Professional Judgment](https://arxiv.org/abs/2606.11678v2)** | 2026-07-11 | <details><summary>This ...</summary>This paper has been withdrawn by the authors because the current version requires substantial revision and further validation before it can be considered a reliable representation of the work</details> |
| **[Beyond Euclidean Clipping: Overcoming Exploration Collapse in LLM RL via Riemannian Isometric Policy Optimization](https://arxiv.org/abs/2607.10169v1)** | 2026-07-11 | ICML 2026 |
| **[EmoStyle: Affective Conditioning of Style-Specialist Experts for Emotional Image Generation](https://arxiv.org/abs/2607.10165v1)** | 2026-07-11 | |
| **[Silent Failures in Quantized LLM Reasoning: A Taxonomy-Based Analysis of Hollow Convergence and Failure Mode Shifts](https://arxiv.org/abs/2607.09999v1)** | 2026-07-10 | <details><summary>7 pag...</summary>7 pages, 3 figures, 6 tables</details> |
| **[Using LLMs to Adjudicate Static-Analysis Alerts with Error Reduction Techniques](https://arxiv.org/abs/2607.09979v1)** | 2026-07-10 | 22 pages |
| **[Hierarchical Chain-of-Thought: Enhancing LLM Reasoning Performance and Efficiency](https://arxiv.org/abs/2604.00130v2)** | 2026-07-10 | |
| **[Practical Source Code Recovery from Binary Functions Using Anchor-Based Retrieval and LLM Reasoning](https://arxiv.org/abs/2607.09452v1)** | 2026-07-10 | 12 pages, 5 figures |
| **[RepLLM: Toward Automatically Reproducing Network Research Results](https://arxiv.org/abs/2509.21074v4)** | 2026-07-09 | <details><summary>SIGCO...</summary>SIGCOMM'26(19 pages, 6 figures, 6 tables)</details> |
| **[Hierarchical Control in Multi-Agent Games: LLM-based Planning and RL Execution](https://arxiv.org/abs/2606.20014v3)** | 2026-07-09 | 12 pages, 9 figures |
| **[TrajAudit: Automated Failure Diagnosis for Agentic Coding Systems](https://arxiv.org/abs/2605.26563v2)** | 2026-07-09 | |
| **[Can We Trust LLM's Logic? Quantifying Uncertainty, Coherence, and Robustness via a Graph-Based Framework](https://arxiv.org/abs/2607.08017v1)** | 2026-07-09 | <details><summary>42 pa...</summary>42 pages, 14 figures, 12 tables</details> |
| **[Fast, Slow, and Tool-augmented Thinking for LLMs: A Review](https://arxiv.org/abs/2508.12265v2)** | 2026-07-08 | <details><summary>The a...</summary>The article has been accepted by Frontiers of Computer Science (FCS), with the DOI: {10.1007/s11704-026-51673-0}</details> |
| **[MILES: Modular Instruction Memory with Learnable Selection for Self-Improving LLM Reasoning](https://arxiv.org/abs/2607.06974v1)** | 2026-07-08 | |
| **[Evaluating SageMath-Augmented LLM Agents for Computational and Experimental Mathematics](https://arxiv.org/abs/2607.06820v1)** | 2026-07-07 | <details><summary>37 pa...</summary>37 pages, 16 figures, accepted to 3rd AI for Math Workshop at ICML 2026</details> |
| **[Knowing When to Quit: A Principled Framework for Dynamic Abstention in LLM Reasoning](https://arxiv.org/abs/2604.18419v6)** | 2026-07-07 | |
| **[Open-Ended Scenario Reasoning for Specialist Model Adaptation](https://arxiv.org/abs/2607.06625v1)** | 2026-07-07 | |
| **[RS-Agent: Automating Remote Sensing Tasks through Intelligent Agent](https://arxiv.org/abs/2406.07089v4)** | 2026-07-07 | |
| **[AgentsCAD: Automated Design for Manufacturing of FDM Parts via Multi-Agent LLM Reasoning and Geometric Feature Recognition](https://arxiv.org/abs/2607.02448v2)** | 2026-07-07 | |
| **[ChargeBD: Character-Aware Heterogeneous Agent Reasoning for Guided Engineering in Battery Development](https://arxiv.org/abs/2606.12057v3)** | 2026-07-06 | |
| **[CAP-CoT: Cycle Adversarial Prompt for Improving Chain of Thoughts in LLM Reasoning](https://arxiv.org/abs/2604.23270v2)** | 2026-07-06 | |
| **[Think Deep, Not Just Long: Measuring LLM Reasoning Effort via Deep-Thinking Tokens](https://arxiv.org/abs/2602.13517v2)** | 2026-07-06 | <details><summary>Accep...</summary>Accepted to ICML 2026</details> |
| **[Interactive Learning for LLM Reasoning](https://arxiv.org/abs/2509.26306v5)** | 2026-07-06 | <details><summary>The c...</summary>The code is available at https://github.com/linhh29/Interactive-Learning-for-LLM-Reasoning</details> |
| **[A Model Can Help Itself: Reward-Free Self-Training for LLM Reasoning](https://arxiv.org/abs/2510.18814v4)** | 2026-07-05 | |
| **[Scalable Semantic Steering of Embedding Projections](https://arxiv.org/abs/2607.03978v1)** | 2026-07-04 | <details><summary>Accep...</summary>Accepted as a short paper at IEEE VIS 2026. 5 pages, 2 figures</details> |
| **[rePIRL: Learn PRM with Inverse RL for LLM Reasoning](https://arxiv.org/abs/2602.07832v3)** | 2026-07-03 | |
| **[LLM-Enhanced Hierarchical Heterogeneous Graph Representation Learning for Malicious Python Package Detection](https://arxiv.org/abs/2607.03350v1)** | 2026-07-03 | |
| **[R$^2$PO: Decoupling Rollout and Inference Policies for LLM Reasoning](https://arxiv.org/abs/2601.11960v3)** | 2026-07-03 | |
| **[OpenSIR: Open-Ended Self-Improving Reasoner](https://arxiv.org/abs/2511.00602v4)** | 2026-07-03 | |

## Chain of Thought
| **Title** | **Date** | **Comment** |
| --- | --- | --- |
| **[LinguistAgent Technical Report: A Reflective Multi-Model Platform for Automated Linguistic Annotation](https://arxiv.org/abs/2602.05493v2)** | 2026-07-21 | |
| **[ResearchArena: Evaluating Sabotage and Monitoring in Automated AI R&D](https://arxiv.org/abs/2607.19321v1)** | 2026-07-21 | 51 pages, 11 figures |
| **[AIM-CoT: Active Information-driven Multimodal Chain-of-Thought for Vision-Language Reasoning](https://arxiv.org/abs/2509.25699v4)** | 2026-07-21 | <details><summary>Accep...</summary>Accepted by ACL 2026 Main Conference. 30 pages, 6 figures</details> |
| **[Cognitive Dual-Process Planning for Autonomous Driving with Structured Scene Knowledge and Verifiable Reasoning-Action Consistency](https://arxiv.org/abs/2607.19194v1)** | 2026-07-21 | <details><summary>12 pa...</summary>12 pages, 7 figures. Zhongyao Yang and Haoyu Li contributed equally to this work</details> |
| **[DAIS: Dependency-Aware Intermediate QA Supervision for Complex Reasoning](https://arxiv.org/abs/2607.19088v1)** | 2026-07-21 | |
| **[Measuring Reward-Seeking via Contrastive Belief Updates](https://arxiv.org/abs/2607.18966v1)** | 2026-07-21 | <details><summary>101 p...</summary>101 pages, 66 figures</details> |
| **[Not All Errors Are Created Equal: ASCoT Addresses Late-Stage Fragility in Efficient LLM Reasoning](https://arxiv.org/abs/2508.05282v6)** | 2026-07-21 | |
| **[Guide, Think, Act: Interactive Embodied Reasoning in Vision-Language-Action Models](https://arxiv.org/abs/2605.13632v2)** | 2026-07-21 | |
| **[CASE: Causal Alignment and Structural Enforcement for Improving Chain-of-Thought Faithfulness](https://arxiv.org/abs/2607.18820v1)** | 2026-07-21 | |
| **[Bounding Boxes to Improve Small Language Model Performance on Vision-Based Grading Tasks](https://arxiv.org/abs/2607.18767v1)** | 2026-07-21 | <details><summary>Accep...</summary>Accepted for 1st Workshop on Small Language Models for Education (SLM4ED '26) at AIED 2026</details> |
| **[Rationale-Guided Knowledge Distillation for Cross-Lingual Stance Detection](https://arxiv.org/abs/2607.18693v1)** | 2026-07-21 | <details><summary>23 pa...</summary>23 pages, 7 figures, 3 tables</details> |
| **[Robust Reasoning Benchmark](https://arxiv.org/abs/2604.08571v3)** | 2026-07-21 | |
| **[LatentMT: Machine Translation with Latent Reasoning](https://arxiv.org/abs/2607.18618v1)** | 2026-07-21 | |
| **[Reasoning Fine-Tuning Induces Persistent Latent Policy States](https://arxiv.org/abs/2607.18532v1)** | 2026-07-20 | <details><summary>Accep...</summary>Accepted at the Conference on Language Modeling (COLM) 2026. 45 pages, including appendices; 24 figures and 12 tables. Code: https://github.com/withmartian/mi-cot</details> |
| **[Measuring and Mitigating Post-hoc Rationalization in Reverse Chain-of-Thought Generation](https://arxiv.org/abs/2602.14469v4)** | 2026-07-20 | ICML 2026 |
| **[Value Leakage: An LLM's Answers Are Silently Shaped by Its Own Values](https://arxiv.org/abs/2607.14345v3)** | 2026-07-20 | |
| **[A Geometric Perspective on Stabilizing Value Conflict Resolution](https://arxiv.org/abs/2607.17946v1)** | 2026-07-20 | <details><summary>Accep...</summary>Accepted to ICML Workshop on High-Dimensional Learning Dynamics</details> |
| **[Benchmarking Resource-Efficient LLMs for Research Topic Ontology Generation in the Biomedical Field](https://arxiv.org/abs/2607.17902v1)** | 2026-07-20 | |
| **[SCA: Segment-Wise CoT Compression with Answer Alignment](https://arxiv.org/abs/2603.07598v2)** | 2026-07-20 | <details><summary>9 pag...</summary>9 pages, 5 figures. This version substantially revises the previous preprint with a new method, updated experiments, and rewritten analysis. Code available at the GitHub project repository https://anonymous.4open.science/r/sca-B666</details> |
| **[ST-Veto: Spatio-Temporal Token Veto for Diffusion MLLMs via Taylor Prediction and Visual Grounding](https://arxiv.org/abs/2607.17884v1)** | 2026-07-20 | ICML 2026 - main |
| **[When a Name Is Not a Name: A Benchmark Dataset and Distilled Reasoning for Culturally Entangled Bangla Homographs in Low-Resource LLMs](https://arxiv.org/abs/2607.17828v1)** | 2026-07-20 | |
| **[MMR-V: What's Left Unsaid? A Benchmark for Multimodal Deep Reasoning in Videos](https://arxiv.org/abs/2506.04141v2)** | 2026-07-20 | <details><summary>Accep...</summary>Accepted at ICLR 2026. Camera-ready version</details> |
| **[Reasoning as a Double-Edged Sword: Architecture and Cross-Stage Robustness in Vision-Language-Action Models](https://arxiv.org/abs/2607.17786v1)** | 2026-07-20 | |
| **[Post-Training Shifts Confidence: A Three-Stage Analysis of How SFT, RL, and OPD Shape CoT Calibration](https://arxiv.org/abs/2607.13753v2)** | 2026-07-20 | |
| **[Coercion and Deception in AI-to-AI Management: An Agentic Benchmark of Unprompted Escalation](https://arxiv.org/abs/2607.15434v2)** | 2026-07-20 | |
| **[LenGuard-GPC: Length Guarding with Guided-Prompt Consistency for Spatial Reasoning Reinforce Learning](https://arxiv.org/abs/2607.17243v1)** | 2026-07-19 | |
| **[Is Your Model Thinking or Just Stagnating? PUMA: Diagnosing Reasoning Pathology via Phase-Momentum Alignment](https://arxiv.org/abs/2607.17188v1)** | 2026-07-19 | |
| **[RoboPIN: Grounded Embodied Reasoning via Pinned Chain-of-Thought](https://arxiv.org/abs/2606.15753v3)** | 2026-07-19 | |
| **[Training Continuous Chain of Thought Models: A Tale of Two Regimes](https://arxiv.org/abs/2607.16972v1)** | 2026-07-18 | <details><summary>Accep...</summary>Accepted to AdaptFM Workshop, ICML 2026</details> |
| **[FUSAR-R1: A Large-Scale Reasoning Model for Intelligent Interpretation of SAR Images](https://arxiv.org/abs/2607.16819v1)** | 2026-07-18 | |
| **[GRASP: Grounded CoT Reasoning with Dual-Stage Optimization for Multimodal Sarcasm Target Identification](https://arxiv.org/abs/2604.08879v2)** | 2026-07-18 | |
| **[Logic-Guided Socially-aware Robot Navigation World Model](https://arxiv.org/abs/2510.23509v3)** | 2026-07-18 | <details><summary>The a...</summary>The authors have decided to withdraw this manuscript due to concerns regarding its current scope, framing, and presentation. Please do not cite this version</details> |
| **[NOWJ@COLIEE 2026: Adaptive Pipelines for Legal Retrieval and Reasoning](https://arxiv.org/abs/2607.16603v1)** | 2026-07-18 | <details><summary>Prese...</summary>Presented at COLIEE 2026</details> |
| **[Visual Access Boundaries in Vision-Language Model Reasoning](https://arxiv.org/abs/2607.12815v2)** | 2026-07-18 | |
| **[From Discussion to Execution: Replicating Buggy and Correct Data Science Code](https://arxiv.org/abs/2607.16569v1)** | 2026-07-18 | <details><summary>accep...</summary>accepted at the 37th IEEE International Symposium on Software Reliability Engineering (ISSRE 2026)</details> |
| **[WeedExpert-R1: Incentivizing Botanical Reasoning in MLLMs with Reinforcement Learning for Precision Weed Grounding](https://arxiv.org/abs/2607.16492v1)** | 2026-07-17 | |
| **[RIMS: Preference Optimization via Smoothed Multi-pair Aggregation for Small-Scale LLM Retrieval-Augmented Generation](https://arxiv.org/abs/2607.16431v1)** | 2026-07-17 | |
| **[Audio-Visual Flamingo: Open Audio-Visual Intelligence for Long and Complex Videos](https://arxiv.org/abs/2607.16107v1)** | 2026-07-17 | <details><summary>Proje...</summary>Project Page: https://avflamingo.pages.dev/</details> |
| **[What Models Express, Suppress, and Resist: Auditing Open-Weight LLMs with Persona Vectors](https://arxiv.org/abs/2607.13162v3)** | 2026-07-17 | |
| **[Latency-Response Theory Model: Evaluating Large Language Models via Response Accuracy and Chain-of-Thought Length](https://arxiv.org/abs/2512.07019v3)** | 2026-07-17 | |
| **[MoT: Modularization-of-Thought Prompting for Effective Code Generation](https://arxiv.org/abs/2503.12483v2)** | 2026-07-17 | |
| **[Length Penalties Make Chain-of-Thought Less Monitorable](https://arxiv.org/abs/2607.09786v2)** | 2026-07-17 | |
| **[LLM-Guided Transportation Hub Capacity Planning with Textual Business Inputs](https://arxiv.org/abs/2607.03651v2)** | 2026-07-17 | |
| **[Better Starts, Better Ends: Bootstrapped Iterative Self-Reasoning Distillation for Compressed Reasoning](https://arxiv.org/abs/2607.15736v1)** | 2026-07-17 | |
| **[Brain-CLIPLM: Semantic Compression for EEG-to-Text Decoding](https://arxiv.org/abs/2604.16370v3)** | 2026-07-17 | |
| **[Video Streaming Thinking: VideoLLMs Can Watch and Think Simultaneously](https://arxiv.org/abs/2603.12262v2)** | 2026-07-17 | <details><summary>Accep...</summary>Accepted by ECCV 2026, project page https://1ranguan.github.io/VST/</details> |
| **[ABot-N1: Toward a General Visual Language Navigation Foundation Model](https://arxiv.org/abs/2607.10383v3)** | 2026-07-17 | |
| **[Neuro-Symbolic AI for LEED compliance: Document-Centric Benchmarking, Deterministic Numeric Checking, and When Multimodal Hurts](https://arxiv.org/abs/2607.15647v1)** | 2026-07-17 | |
| **[RecGPT-V3 Technical Report](https://arxiv.org/abs/2607.15591v1)** | 2026-07-17 | Technique Report |

## LLM Interpretability
| **Title** | **Date** | **Comment** |
| --- | --- | --- |
| **[Understanding Interpretation Difficulty in Harmful Online Communication: Insights from Cybercrime Communities](https://arxiv.org/abs/2607.07277v1)** | 2026-07-08 | |
| **[NeuraDock Visual Cognitive Load Agent Tutorial: A Quality-Gated Open-Source EEG Workflow for Alpha Dynamics and Real-Time Applications](https://arxiv.org/abs/2606.26518v1)** | 2026-06-25 | 22 pages, 10 figures |
| **[Don't Go Breaking My LLM: The Impact of Pruning Attention Layers on Explanation Faithfulness and Confidence Calibration](https://arxiv.org/abs/2606.24970v1)** | 2026-06-23 | Accepted at TMLR |
| **[LLM-Based Generalizable Hierarchical Task Planning and Execution for Heterogeneous Robot Teams with Event-Driven Replanning](https://arxiv.org/abs/2511.22354v2)** | 2026-06-18 | <details><summary>full ...</summary>full version of this short paper is accepted at Frontiers in Robotics and AI Journal</details> |
| **[ICA Lens: Interpreting Language Models Without Training Another Dictionary](https://arxiv.org/abs/2606.11722v1)** | 2026-06-10 | Ongoing Project |
| **[Language-Driven Cost Optimization for Autonomous Driving](https://arxiv.org/abs/2606.10974v1)** | 2026-06-09 | <details><summary>Paper...</summary>Paper accepted at IEEE Intelligent Transportation Systems Conference (ITSC) 2026</details> |
| **[ReflectiChain: Epistemic Grounding in LLM-Driven World Models for Supply Chain Resilience](https://arxiv.org/abs/2606.10359v1)** | 2026-06-09 | |
| **[MotionGPT-2: A General-Purpose Motion-Language Model for Motion Generation and Understanding](https://arxiv.org/abs/2410.21747v2)** | 2026-06-08 | |
| **[Analyzing the Correlation Between Hallucinations and Knowledge Conflicts in Large Language Models](https://arxiv.org/abs/2606.08705v1)** | 2026-06-07 | |
| **[Thinking Through Signs: PEEL as a Semiotic Scaffolding for Epistemically Accountable AI-Enabled Research](https://arxiv.org/abs/2606.04152v1)** | 2026-06-02 | 10 pages, 5 figuras |
| **[Unveiling the Limits of Large Language Models in Inferring Pragmatic Meaning from Non-Verbal Responses](https://arxiv.org/abs/2606.01845v1)** | 2026-06-01 | |
| **[The Limits of LLM Forecasting: Parametric Knowledge Gaps Across Conflict Zones](https://arxiv.org/abs/2607.00018v1)** | 2026-05-29 | |
| **[Hallucination Behavior in Multimodal LLMs Across Agricultural Image Interpretation and Generation Tasks](https://arxiv.org/abs/2605.27595v1)** | 2026-05-26 | |
| **[How Much Structure Do LLMs Need? Evaluating LLMs for Bibliometric Cluster Description](https://arxiv.org/abs/2605.24351v1)** | 2026-05-23 | |
| **[Improving Labeling Consistency with Detailed Constitutional Definitions and AI-Driven Evaluation](https://arxiv.org/abs/2605.24247v1)** | 2026-05-22 | <details><summary>Under...</summary>Under review at ACL Rolling Review (ARR), May 2026 cycle. Also available at https://doi.org/10.5281/zenodo.20125267</details> |
| **[Patch-Effect Graph Kernels for LLM Interpretability](https://arxiv.org/abs/2605.06480v1)** | 2026-05-07 | |
| **[LLM-ADAM: A Generalizable LLM Agent Framework for Pre-Print Anomaly Detection in Additive Manufacturing](https://arxiv.org/abs/2605.03328v1)** | 2026-05-05 | 24 pages, 10 figures |
| **[From Research Question to Scientific Workflow: Leveraging Agentic AI for Science Automation](https://arxiv.org/abs/2604.21910v1)** | 2026-04-23 | |
| **[Agentic AI-Enabled Framework for Thermal Comfort and Building Energy Assessment in Tropical Urban Neighborhoods](https://arxiv.org/abs/2604.21787v1)** | 2026-04-23 | <details><summary>Accep...</summary>Accepted at IAQVEC 2026</details> |
| **[Learning to Draw ASCII Improves Spatial Reasoning in Language Models](https://arxiv.org/abs/2604.14641v1)** | 2026-04-16 | |
| **[Arbitration Failure, Not Perceptual Blindness: How Vision-Language Models Resolve Visual-Linguistic Conflicts](https://arxiv.org/abs/2604.09364v2)** | 2026-04-13 | |
| **[Heterogeneous Graph Importance Scoring and Clustering with Automated LLM-based Interpretation](https://arxiv.org/abs/2605.02919v1)** | 2026-04-09 | <details><summary>26 pa...</summary>26 pages, 11 figures, 8 tables</details> |
| **[Runtime Execution Traces Guided Automated Program Repair with Multi-Agent Debate](https://arxiv.org/abs/2604.02647v1)** | 2026-04-03 | <details><summary>12 pa...</summary>12 pages, 4 figures, 8 tables</details> |
| **[Discovering Decoupled Functional Modules in Large Language Models](https://arxiv.org/abs/2603.17823v1)** | 2026-03-18 | AAAI-26 Oral |
| **[LaMoGen: Language to Motion Generation Through LLM-Guided Symbolic Inference](https://arxiv.org/abs/2603.11605v1)** | 2026-03-12 | <details><summary>Accep...</summary>Accepted by CVPR 2026. Supplementary material included. Project page: https://jjkislele.github.io/LaMoGen/</details> |
| **[Pneuma-Seeker: A Relational Reification Mechanism to Align AI Agents with Human Work over Relational Data](https://arxiv.org/abs/2603.10747v1)** | 2026-03-11 | |
| **[PhysLLM: Harnessing Large Language Models for Cross-Modal Remote Physiological Sensing](https://arxiv.org/abs/2505.03621v2)** | 2026-03-05 | <details><summary>Accep...</summary>Accepted by International Conference on Learning Representations (ICLR) 2026</details> |
| **[Cache What Lasts: Token Retention for Memory-Bounded KV Cache in LLMs](https://arxiv.org/abs/2512.03324v2)** | 2026-03-01 | |
| **[Sparse Shift Autoencoders for Identifying Concepts from Large Language Model Activations](https://arxiv.org/abs/2502.12179v2)** | 2026-02-27 | 27 pages, 9 figures |
| **[Co-Disclosing the Computer: LLM-Mediated Computing through Reflective Conversation](https://arxiv.org/abs/2604.09586v1)** | 2026-02-27 | CHI'26 |
| **[Neuro-Symbolic Control with Large Language Models for Language-Guided Spatial Tasks](https://arxiv.org/abs/2512.17321v2)** | 2026-02-21 | |
| **[Stroke Lesions as a Rosetta Stone for Language Model Interpretability](https://arxiv.org/abs/2602.04074v1)** | 2026-02-03 | 45 pages, 17 figures |
| **[Uncertainty-Aware Gradient Signal-to-Noise Data Selection for Instruction Tuning](https://arxiv.org/abs/2601.13697v1)** | 2026-01-20 | Preprint |
| **[A Shared Geometry of Difficulty in Multilingual Language Models](https://arxiv.org/abs/2601.12731v1)** | 2026-01-19 | |
| **[Causal-SAM-LLM: Large Language Models as Causal Reasoners for Robust Medical Segmentation](https://arxiv.org/abs/2507.03585v2)** | 2026-01-16 | <details><summary>Accep...</summary>Accepted by IEEE ICASSP 2026</details> |
| **[Self-reflection in Automated Qualitative Coding: Improving Text Annotation through Secondary LLM Critique](https://arxiv.org/abs/2601.09905v1)** | 2026-01-14 | |
| **[Word Synchronization Challenge: A Benchmark for Word Association Responses for Large Language Models](https://arxiv.org/abs/2502.08312v2)** | 2026-01-14 | |
| **[Can LLMs interpret figurative language as humans do?: surface-level vs representational similarity](https://arxiv.org/abs/2601.09041v1)** | 2026-01-14 | 17 pages, 5 figures |
| **[Neuro-Symbolic Compliance: Integrating LLMs and SMT Solvers for Automated Financial Legal Analysis](https://arxiv.org/abs/2601.06181v1)** | 2026-01-07 | <details><summary>10 pa...</summary>10 pages, 6 tables, 3 figures, accepted by the 2nd ACM AIware Conference</details> |
| **[LLM Interpretability with Identifiable Temporal-Instantaneous Representation](https://arxiv.org/abs/2509.23323v2)** | 2026-01-02 | NeurIPS 2025 |
| **[Trust Me, I Know This Function: Hijacking LLM Static Analysis using Bias](https://arxiv.org/abs/2508.17361v2)** | 2025-12-18 | |
| **[Verification-Guided Context Optimization for Tool Calling via Hierarchical LLMs-as-Editors](https://arxiv.org/abs/2512.13860v1)** | 2025-12-15 | <details><summary>Accep...</summary>Accepted by AAAI 2026 Workshop on Agentic AI Benchmarks and Applications for Enterprise Tasks</details> |
| **[Visualizing token importance for black-box language models](https://arxiv.org/abs/2512.11573v1)** | 2025-12-12 | |
| **[SpeechIQ: Speech-Agentic Intelligence Quotient Across Cognitive Levels in Voice Understanding by Large Language Models](https://arxiv.org/abs/2507.19361v2)** | 2025-12-01 | <details><summary>ACL 2...</summary>ACL 2025 main. Our Speech-IQ leaderboard is hosted at huggingface.co/spaces/nvidia/Speech-IQ-leaderboard. Speech-IQ Calculator: https://github.com/YukinoWan/SpeechIQ</details> |
| **[For Those Who May Find Themselves on the Red Team](https://arxiv.org/abs/2511.18499v1)** | 2025-11-23 | |
| **[HSKBenchmark: Modeling and Benchmarking Chinese Second Language Acquisition in Large Language Models through Curriculum Tuning](https://arxiv.org/abs/2511.15574v1)** | 2025-11-19 | <details><summary>Accep...</summary>Accepted by AAAI-2026</details> |
| **[KnowThyself: An Agentic Assistant for LLM Interpretability](https://arxiv.org/abs/2511.03878v1)** | 2025-11-05 | <details><summary>5 pag...</summary>5 pages, 1 figure, Accepted for publication at the Demonstration Track of the 40th AAAI Conference on Artificial Intelligence (AAAI 26)</details> |
| **[Imperfect Language, Artificial Intelligence, and the Human Mind: An Interdisciplinary Approach to Linguistic Errors in Native Spanish Speakers](https://arxiv.org/abs/2511.01615v1)** | 2025-11-03 | 12 pages, 3 figures |
| **[To Sink or Not to Sink: Visual Information Pathways in Large Vision-Language Models](https://arxiv.org/abs/2510.08510v1)** | 2025-10-09 | <details><summary>Prepr...</summary>Preprint. Project page: https://davidhalladay.github.io/diysink_demo</details> |
| **[Automated Repair of Ambiguous Problem Descriptions for LLM-Based Code Generation](https://arxiv.org/abs/2505.07270v3)** | 2025-09-24 | |

## Explainable AI
| **Title** | **Date** | **Comment** |
| --- | --- | --- |
| **[LLM-Grounded Explainable AI for Supply Chain Risk Early Warning via Temporal Graph Attention Networks](https://arxiv.org/abs/2603.04818v3)** | 2026-07-21 | |
| **[Automated Data Engineering and Feature Selection for the Case Study of Warpage Detection in Fused Deposition Modeling](https://arxiv.org/abs/2607.18515v1)** | 2026-07-20 | |
| **[Explaining and Tuning Transformer-based LLMs in Arithmetic Tasks with Human Strategies](https://arxiv.org/abs/2607.17166v1)** | 2026-07-19 | |
| **[FST.ai 2.5: Explainable and Uncertainty-Aware AI for Olympic and Para-Taekwondo Decision Support, Athlete Digital Twins, and Federation-Scale Analytics](https://arxiv.org/abs/2607.16597v1)** | 2026-07-18 | 14 pages, 4 figures |
| **[FAIR_XAI: Improving Multimodal Foundation Model Fairness via Explainability for Wellbeing Assessment](https://arxiv.org/abs/2604.23786v2)** | 2026-07-17 | 11 pages, 4 figures |
| **[VeriX-Anon: A Multi-Layered Framework for Mathematically Verifiable Outsourced Target-Driven Data Anonymization](https://arxiv.org/abs/2604.12431v2)** | 2026-07-17 | <details><summary>v2: r...</summary>v2: revised after peer review. Evaluation expanded from 3 to 7 datasets, per-dataset Wasserstein-threshold calibration added, effect-size CIs and cross-dataset statistics reported, and analytical zk-SNARK/MPC/TEE baselines added. Minor errors corrected</details> |
| **[Scaling Time Series Classification via XAI-Driven Data Reduction](https://arxiv.org/abs/2607.15774v1)** | 2026-07-17 | <details><summary>Accep...</summary>Accepted for AALTD workshop at ECML-PKDD 2026</details> |
| **[Automated identification of Ichneumonoidea wasps via YOLO-based deep learning: Integrating HiresCam for Explainable AI](https://arxiv.org/abs/2603.16351v2)** | 2026-07-16 | 15 pages, 20 figures |
| **[Explaining Process Control Optimisation Recommendations via GradientSHAP and Implicit Differentiation](https://arxiv.org/abs/2607.14970v1)** | 2026-07-16 | |
| **[Diagnosing and Mitigating Domain Shift in Permission-Based Android Malware Detection](https://arxiv.org/abs/2605.09028v3)** | 2026-07-15 | |
| **["Trust Junk" Leads to Unjustified Support for Highly Discriminatory Predictive Models](https://arxiv.org/abs/2607.14152v1)** | 2026-07-14 | |
| **[MLPTR-CC: Multi-label Pathology Test Recommendation using Classifier Chains and SHAP](https://arxiv.org/abs/2607.08299v2)** | 2026-07-14 | |
| **[Atomic Units of X: The Compression Layer of Intelligence](https://arxiv.org/abs/2607.12634v1)** | 2026-07-14 | |
| **[Evaluating RE Practices for Explainability: Synthesizing Insights from Daimler Truck into an Explainable RE Framework Proposal](https://arxiv.org/abs/2607.11771v1)** | 2026-07-13 | |
| **[Detecting Explanatory Insufficiency in Learned Representations: A Framework for Representational Vigilance](https://arxiv.org/abs/2606.13172v2)** | 2026-07-13 | <details><summary>25 pa...</summary>25 pages, 2 figures, no tables, 16 references. Conceptual and methodological framework for monitoring representational adequacy and detecting explanatory insufficiency in learned representations</details> |
| **[ConceptSMILE: Auditing the Trustworthiness of Concept-Based Explainable AI](https://arxiv.org/abs/2607.09649v1)** | 2026-07-10 | |
| **[Knowledge Graphs and Explainable AI as Complementary Resources for Urban Mining](https://arxiv.org/abs/2607.09578v1)** | 2026-07-10 | <details><summary>Accep...</summary>Accepted for presentation at the AISE Workshop @ IJCAI-ECAI 2026</details> |
| **[All Explanations are Wrong, But Many Are Useful: Exploring the Rashomon Explanation Set with Large Language Models](https://arxiv.org/abs/2607.09502v1)** | 2026-07-10 | |
| **[Application of machine learning to monster level prediction in tabletop RPG game design](https://arxiv.org/abs/2607.09196v1)** | 2026-07-10 | |
| **[Unlearning to Protect: A Distilled Reinforcement Learning Framework with Privacy-Preserving Feature Unlearning and XAI for IoT Security](https://arxiv.org/abs/2607.07635v2)** | 2026-07-09 | |
| **[The Contribution of XAI for the Safe Development and Certification of AI: An Expert-Based Analysis](https://arxiv.org/abs/2408.02379v2)** | 2026-07-09 | <details><summary>This ...</summary>This preprint has not undergone peer review or any post-submission improvements or corrections. The Version of Record of this contribution is published in Artificial Intelligence in HCI (HCII 2026), Lecture Notes in Computer Science, vol. 16745, and is available online at https://doi.org/10.1007/978-3-032-30849-8_13</details> |
| **[Mechanistic Interpretability for Neural Networks: Circuits, Sparse Features and Symbolic Reasoning](https://arxiv.org/abs/2607.07316v1)** | 2026-07-08 | 20 pages |
| **[Why Fake ? Unveiling the Semantic Vocabulary of Deepfake Detectors](https://arxiv.org/abs/2607.07216v1)** | 2026-07-08 | <details><summary>Accep...</summary>Accepted at CVPRW 2026</details> |
| **[Large Language Models (LLMs) and Generative AI in Cybersecurity and Privacy: A Survey of Dual-Use Risks, AI-Generated Malware, Explainability, and Defensive Strategies](https://arxiv.org/abs/2607.06963v1)** | 2026-07-08 | <details><summary>Invit...</summary>Invited survey paper. 10 pages, 5 figures, 2 tables</details> |
| **[Reduced NEXI protocol for the quantification of human gray matter microstructure on the Connectome 2.0 scanner](https://arxiv.org/abs/2509.09513v3)** | 2026-07-07 | <details><summary>Submi...</summary>Submitted to Imaging Neuroscience. This all-in-one version includes supplementary materials. 34 pages, 145 figures, 4 tables</details> |
| **[Explainable embeddings with Distance Explainer](https://arxiv.org/abs/2505.15516v3)** | 2026-07-07 | <details><summary>21 pa...</summary>21 pages, 12 figures. Accepted to the 4th World Conference on eXplainable Artificial Intelligence. Method implementation: https://research-software-directory.org/software/distance-explainer</details> |
| **[Measuring What Matters: A Unified Evaluation Framework for GNN Explainability](https://arxiv.org/abs/2607.04600v1)** | 2026-07-06 | |
| **[Dynamic Interest Rate Discovery in Decentralized Finance: A Reverse Kelly Automated Market Maker for Risk-Adjusted Lending](https://arxiv.org/abs/2607.04178v1)** | 2026-07-05 | |
| **[Explainable AI for Screening Abuse-Related Trauma in Bangladeshi Children: A Training-Free Multimodal Framework Evaluated on Noise-Aware Synthetic Data](https://arxiv.org/abs/2607.04010v1)** | 2026-07-04 | 6 pages, 5 figures |
| **[Activation-Deactivation: A General Framework for Robust Post-hoc Explainable AI](https://arxiv.org/abs/2510.01038v2)** | 2026-07-04 | <details><summary>Prepr...</summary>Preprint: Under Review; Updated experiments & Figures</details> |
| **[Better Together? The Role of Explanations in Supporting Novices in Individual and Collective Deliberations about AI](https://arxiv.org/abs/2411.11449v3)** | 2026-07-03 | <details><summary>30 pa...</summary>30 pages main text, 8 figures, 4 tables. Supplementary material is included in the appendix</details> |
| **[Algebraic Model Counting for Global Analysis of Optimal Decision Trees](https://arxiv.org/abs/2607.02069v1)** | 2026-07-02 | <details><summary>Proc....</summary>Proc. Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD 2026), LNCS, Naples, Italy, 7-11 September 2026</details> |
| **[Quantum-Inspired Vision: Leveraging Wave-Particle Duality for Low-Illumination Enhancement](https://arxiv.org/abs/2607.01731v1)** | 2026-07-02 | |
| **[Reducing Labeling Effort in Architecture Technical Debt Detection through Active Learning and Explainable AI](https://arxiv.org/abs/2603.02944v2)** | 2026-07-01 | <details><summary>Accep...</summary>Accepted for publication in Empirical Software Engineering (EMSE) Journal, 2026</details> |
| **[PACE: A Neuro-Symbolic Framework for Plausible and Actionable Counterfactual Explanations](https://arxiv.org/abs/2607.01306v1)** | 2026-07-01 | |
| **[CPG-PAD: Concept-Informed Prompts Guided Presentation Attack Detection](https://arxiv.org/abs/2607.01303v1)** | 2026-07-01 | <details><summary>Accep...</summary>Accepted by IEEE Transactions on Information Forensics & Security (TIFS)</details> |
| **[Understanding Large Language Models](https://arxiv.org/abs/2607.01006v1)** | 2026-07-01 | 25 pages, 1 figure |
| **[Explainable AI for Cancer Drug Response Prediction: Beyond Univariate Feature Attributions](https://arxiv.org/abs/2607.00931v1)** | 2026-07-01 | |
| **[FedXDS: Leveraging Model Attribution Methods to counteract Data Heterogeneity in Federated Learning](https://arxiv.org/abs/2606.31742v1)** | 2026-06-30 | |
| **[Scientific Explanations in Health Sciences: Causality, Trust, and Epistemic Adequacy](https://arxiv.org/abs/2606.31616v1)** | 2026-06-30 | |
| **[Explainable AI in Speaker Recognition -- Making Latent Representations Understandable](https://arxiv.org/abs/2604.23354v3)** | 2026-06-29 | A working paper |
| **[Feature-level Interaction Explanations in Multimodal Transformers](https://arxiv.org/abs/2603.13326v2)** | 2026-06-28 | |
| **[Understanding LLM Intervention Explanations in Multi-Party Human-Robot Interaction](https://arxiv.org/abs/2606.29460v1)** | 2026-06-28 | <details><summary>Accep...</summary>Accepted for 2026 36th IEEE International Conference on Robot and Human Interactive Communication</details> |
| **[Low-cost concept-based localized explanations: How far can we get with training-free approaches?](https://arxiv.org/abs/2606.29069v1)** | 2026-06-27 | <details><summary>6 pag...</summary>6 pages, 2 figures, 4 tables. Accepted at the 2026 IEEE International Conference on Artificial Intelligence (CAI), 8-10 May 2026, Granada, Spain. Code: https://github.com/darianfgUgr/CoNa</details> |

## Mechanistic Interpretability
| **Title** | **Date** | **Comment** |
| --- | --- | --- |
| **[CircuitKIT : Circuit Discovery, Evaluation, and Application Toolkit for Mechanistic Interpretability](https://arxiv.org/abs/2607.19317v1)** | 2026-07-21 | |
| **[CLT-Forge: A Scalable Library for Cross-Layer Transcoders and Attribution Graphs](https://arxiv.org/abs/2603.21014v2)** | 2026-07-21 | <details><summary>9 pag...</summary>9 pages, 7 figures, 1 table. Code: https://github.com/LLM-Interp/CLT-Forge. Demonstration video: https://youtu.be/6ptrrLawTl8</details> |
| **[EvalSafetyGap: A Hybrid Survey and Conceptual Framework for LLM Evaluation-Safety Failures](https://arxiv.org/abs/2606.30219v3)** | 2026-07-21 | <details><summary>This ...</summary>This manuscript is a 80-page hybrid survey and conceptual framework on LLM evaluation and AI-safety failures. It includes 8 figures and multiple evidence-synthesis tables, covering literature from 2018 to 2026. The paper introduces the EvalSafetyGap framework and reports a structured audit of 10 LLMs. It is submitted as a review/survey article and is not currently under consideration elsewhere</details> |
| **[Breaking the Block: Preserving Data Continuity to Train Superior SAEs for Instruct Models](https://arxiv.org/abs/2506.07691v2)** | 2026-07-20 | |
| **[Measuring Monosemanticity in Sparse Autoencoders via Latent Activation Coherence](https://arxiv.org/abs/2607.17770v1)** | 2026-07-20 | <details><summary>This ...</summary>This is a preprint version. A shorter version of this paper has been accepted for presentation and publication in the post-workshop proceedings of the 8th International Workshop on eXplainable Knowledge Discovery in Data Mining (XKDD 2026), co-located with ECML PKDD 2026. The appendix is included only in this preprint and is not part of the peer-reviewed proceedings paper</details> |
| **[Every Component is a Lookup: Token Attribution and Composition from a Single Decomposition](https://arxiv.org/abs/2605.23393v2)** | 2026-07-19 | |
| **[What Do They See? Interpreting Complex Road Scenarios Through the Eyes of Vision-Language-Action Models for Safe and Trustworthy Autonomous Vehicle Learning](https://arxiv.org/abs/2607.16938v1)** | 2026-07-18 | |
| **[Are Arithmetic Heuristic Neurons Form-Invariant? A Mechanistic Analysis of Symbols, Text, and Code in LLMs](https://arxiv.org/abs/2607.16693v1)** | 2026-07-18 | Under Review |
| **[Mechanistic Interpretability of Cognitive Complexity in LLMs via Linear Probing using Bloom's Taxonomy](https://arxiv.org/abs/2602.17229v2)** | 2026-07-17 | <details><summary>Prepr...</summary>Preprint. Under review</details> |
| **[Analysing Moral Bias in Finetuned LLMs through Mechanistic Interpretability](https://arxiv.org/abs/2510.12229v3)** | 2026-07-17 | <details><summary>Prepr...</summary>Preprint. Under review</details> |
| **[Hide and Seek in Embedding Space: Geometry-based Steganography and Detection in Large Language Models](https://arxiv.org/abs/2601.22818v2)** | 2026-07-17 | |
| **[Steering Robustness into World Action Models via Mechanistic Interpretability and Optimal Control](https://arxiv.org/abs/2607.14943v1)** | 2026-07-16 | |
| **[Transcoders for Investigating Deception in Language Models](https://arxiv.org/abs/2607.14791v1)** | 2026-07-16 | |
| **[Tracing LLM Behavior to the Training Data with Empirical Next-Token Distributions](https://arxiv.org/abs/2607.14306v1)** | 2026-07-15 | |
| **[From Observation to Insight: Mechanistic World Models and the Quest for Autonomous Discovery](https://arxiv.org/abs/2607.12474v2)** | 2026-07-15 | |
| **[Hierarchical Latent Structures in Data Generation Process Unify Mechanistic Phenomena across Scale](https://arxiv.org/abs/2603.06592v2)** | 2026-07-14 | |
| **[Same Compression Principle, Different Geometry: Rate-Distortion Signatures Dissociate Biological and Artificial Visual Systems](https://arxiv.org/abs/2603.01568v2)** | 2026-07-13 | |
| **[Inside the Unfair Judge: A Mechanistic Interpretability Account of LLM-as-Judge Bias](https://arxiv.org/abs/2607.11871v1)** | 2026-07-13 | <details><summary>58 pa...</summary>58 pages, 13 figures, 30 tables; project page: https://xzx34.github.io/unfair-judge/</details> |
| **[Emergent Hierarchical Monosemantic Neurons from the Group-Contrastive Forward-Forward Algorithm](https://arxiv.org/abs/2607.16295v1)** | 2026-07-13 | |
| **[Which Neurons Detect Malicious Code? A Probing Study of LLM Security Knowledge](https://arxiv.org/abs/2607.10221v1)** | 2026-07-11 | <details><summary>The p...</summary>The paper has been peer reviewed and accepted for publication in the 20th International Symposium on Empirical Software Engineering and Measurement (ESEM 2026)</details> |
| **[MLPs are Hebbians: Constructing Efficient Fact-Storing MLPs for Transformers](https://arxiv.org/abs/2607.10034v1)** | 2026-07-10 | |
| **[XAI and Statistical Analysis for Reliable Intrusion Detection in the UAVIDS-2025 Dataset: From Tree to Hybrid and Tabular DNN Ensembles](https://arxiv.org/abs/2605.13922v2)** | 2026-07-10 | <details><summary>Accep...</summary>Accepted at IEEE CITS 2026, Greece</details> |
| **[When Structured Sparse Autoencoders Learn Consistent Concepts Across Modalities](https://arxiv.org/abs/2607.08605v1)** | 2026-07-09 | |
| **[Cross-seed explainability using Procrustes-conditioned Joint End-to-end Top-K Sparse Autoencoders](https://arxiv.org/abs/2607.08499v1)** | 2026-07-09 | <details><summary>17 pa...</summary>17 pages, 4 figures, 6 tables</details> |
| **[Towards Isolated Interventions via Almost Orthogonal Features in Language Models](https://arxiv.org/abs/2602.04718v2)** | 2026-07-09 | <details><summary>Accep...</summary>Accepted as a conference paper at the Conference on Language Modeling (COLM) 2026</details> |
| **[Certified Interventional Fidelity: Anytime-Valid, Adaptive Evaluation of Causal Claims in Mechanistic Interpretability](https://arxiv.org/abs/2607.08349v1)** | 2026-07-09 | <details><summary>Accep...</summary>Accepted at UAI 2026 (Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence). Code: https://github.com/AsiaeeLab/certified-interventional-fidelity</details> |
| **[Diagnosing Shape-Prior Shortcuts in Long-Range Single-Shot Fringe Projection Profilometry](https://arxiv.org/abs/2606.17093v2)** | 2026-07-09 | 21 pages, 13 figures |
| **[Mechanistic Interpretability of LLM Jailbreaks via Internal Attribution Graphs](https://arxiv.org/abs/2607.07903v1)** | 2026-07-08 | |
| **[Temporal Preference Concepts and their Functions in a Large Language Model](https://arxiv.org/abs/2606.05194v2)** | 2026-07-08 | |
| **[How Learning Dynamics Drive Adversarially Robust Generalization?](https://arxiv.org/abs/2410.07719v4)** | 2026-07-08 | <details><summary>Accep...</summary>Accepted at the 42nd Conference on Uncertainty in Artificial Intelligence (UAI 2026)</details> |
| **[Mechanistic Interpretability for Neural Networks: Circuits, Sparse Features and Symbolic Reasoning](https://arxiv.org/abs/2607.07316v1)** | 2026-07-08 | 20 pages |
| **[Latent Programming Horizons in Coding Agents](https://arxiv.org/abs/2607.05188v1)** | 2026-07-06 | |
| **[Beyond the Black Box: Interpretability of Agentic AI Tool Use](https://arxiv.org/abs/2605.06890v4)** | 2026-07-05 | <details><summary>12 pa...</summary>12 pages, 4 figures, 17 tables</details> |
| **[Interpretability and Generalization Bounds for Learning Spatial Physics](https://arxiv.org/abs/2506.15199v4)** | 2026-07-05 | <details><summary>To ap...</summary>To appear in ICML 2026. 18 pages, 13 figures</details> |
| **[Cultural Binding Heads in Language Models](https://arxiv.org/abs/2605.28543v2)** | 2026-07-04 | |
| **[Using Mechanistic Interpretability to Craft Adversarial Attacks against Large Language Models](https://arxiv.org/abs/2503.06269v3)** | 2026-07-03 | |
| **[Individual Parameters in Weight-Sparse Transformers Appear Interpretable](https://arxiv.org/abs/2607.02964v1)** | 2026-07-03 | <details><summary>20 pa...</summary>20 pages, 19 figures, 3 tables. Project website: https://weightpedia.org/individual-parameters-in-sparse-transformers/</details> |
| **[MatPhaseBench: A Semantics-Guided Benchmark for Materials Phase Diagrams Understanding](https://arxiv.org/abs/2607.02934v1)** | 2026-07-03 | |
| **[Induction Heads Interpolate N-Grams](https://arxiv.org/abs/2607.02800v1)** | 2026-07-02 | <details><summary>Publi...</summary>Published as a conference paper at ICML 2026. OpenReview: https://openreview.net/forum?id=BSY7jhBxM1</details> |
| **[Towards Robustness against Typographic Attack with Training-free Concept Localization](https://arxiv.org/abs/2607.02494v1)** | 2026-07-02 | <details><summary>15 pa...</summary>15 pages main text, provisionally accepted to ECCV 2026</details> |
| **[Conditional Co-Ablation: Recovering Self-Repair Backups in Transformer Circuits](https://arxiv.org/abs/2607.01940v1)** | 2026-07-02 | |
| **[Shared Semantics, Divergent Mechanisms: Unsupervised Feature Discovery by Aligning Semantics and Mechanisms](https://arxiv.org/abs/2606.08236v2)** | 2026-07-02 | <details><summary>40 pa...</summary>40 pages; accepted as an ICML 2026 Spotlight; project page: https://merenova.github.io/distribution-level-feature-discovery/</details> |
| **[Expander Sparse Autoencoders: Parameter-Efficient Dictionaries for Mechanistic Interpretability](https://arxiv.org/abs/2607.01799v1)** | 2026-07-02 | |
| **[Mechanistic Interpretability and Causal Feature Steering of Neural Quantum States via Sparse Autoencoders](https://arxiv.org/abs/2607.01336v1)** | 2026-07-01 | <details><summary>15 pa...</summary>15 pages, 7 figures. Comments welcome!</details> |
| **[Muon as a Residual Connection](https://arxiv.org/abs/2607.01124v1)** | 2026-07-01 | |
| **[Interpreting Global Perturbation Robustness of Image Models using Axiomatic Spectral Importance Decomposition](https://arxiv.org/abs/2408.01139v4)** | 2026-07-01 | <details><summary>Accep...</summary>Accepted by Transactions on Machine Learning Research (TMLR 2024)</details> |
| **[MetaOthello: A Controlled Study of Multiple World Models in Transformers](https://arxiv.org/abs/2602.23164v2)** | 2026-07-01 | <details><summary>Camer...</summary>Camera-ready version. Accepted to the 43rd International Conference on Machine Learning (ICML 2026)</details> |
| **[Representation as a Bottleneck for Mechanistic Interpretability: The Manifestation Unit Protocol](https://arxiv.org/abs/2607.00089v1)** | 2026-06-30 | <details><summary>65 pa...</summary>65 pages. Interactive demos: https://manifestation-xai.github.io/manifestation-transformers/ , https://manifestation-xai.github.io/manifestation-cnn</details> |
| **[Surrogate Fidelity: When Can Open LLMs Explain Closed Ones?](https://arxiv.org/abs/2606.32008v1)** | 2026-06-30 | |

Title	Date	Comment
They'll Verify. They Just Won't Act. How Authority Framing and Laundered Code Turn a Trusted Agentic CI/CD Pipeline Into an Attack Surface	2026-07-21	9 pag... 9 pages. Dataset and reproduction code: https://github.com/senthex-security/senthex-research
Reasoning Error from Known Fact: Step-Level Self-Consistency Group Relative Policy Optimization for LLM	2026-07-21
Tracing the Shadows: Automatic Tracking and Analysis of Crypto Money Laundering via Transaction Semantic Analysis	2026-07-21
Not All Errors Are Created Equal: ASCoT Addresses Late-Stage Fragility in Efficient LLM Reasoning	2026-07-21
LLM-Grounded Explainable AI for Supply Chain Risk Early Warning via Temporal Graph Attention Networks	2026-07-21
QuArch: A Benchmark for Evaluating LLM Reasoning in Computer Architecture	2026-07-21
Oracle Gap and Signal Fidelity: A Fixed-Pool Diagnostic for Test-Time Collaboration	2026-07-20	13 pa... 13 pages, 2 figures, 8 tables. Code: https://github.com/AmGarfield/OracleGap
LLMs as a Jury: Cross-Model Consensus Can Outperform Process Reward Models for LLM Reasoning	2026-07-20
AIGB-R1: Self-Evolving Generative Auto-Bidding via Hierarchical Planner-Executor Optimization	2026-07-19
Debate-on-Graph: Reliable and Adaptive Reasoning of Large Language Model on Uncertain Knowledge Graph	2026-07-19	18 pa... 18 pages, Accepted by ECML-PKDD 2026
Constrained Path Reasoning: Measuring When Committed Stages Earn Their Cost	2026-07-19
Prismatic Synthesis: Gradient-based Data Diversification Boosts Generalization in LLM Reasoning	2026-07-19
Logic-Guided Socially-aware Robot Navigation World Model	2026-07-18	The a... The authors have decided to withdraw this manuscript due to concerns regarding its current scope, framing, and presentation. Please do not cite this version
NOWJ@COLIEE 2026: Adaptive Pipelines for Legal Retrieval and Reasoning	2026-07-18	Prese... Presented at COLIEE 2026
Video Streaming Thinking: VideoLLMs Can Watch and Think Simultaneously	2026-07-17	Accep... Accepted by ECCV 2026, project page https://1ranguan.github.io/VST/
REST: Receding Horizon Explorative Steiner Tree for Zero-Shot Object-Goal Navigation	2026-07-16	Accepted to IROS'26
MARS: Multi-hop Adaptive Retrieval and SPARQL Generation for KGQA	2026-07-16	EKAW ... EKAW 2026 (https://ekaw2026.di.unito.it/accepted-posters-and-demos)
ProfMalPlus: Agent-Coordinated Detection of Malicious NPM Packages via Static-Dynamic Analysis Synergy	2026-07-15
Policy of Thoughts: Scaling Test-Time Training for LLM Reasoning via Online Policy Evolution	2026-07-15	Under... Under Review, preprint
Assert, don't describe: Linguistic features that shift LLM reasoning about animal welfare	2026-07-12
Cognitive Episodes in LLM Reasoning Traces Enable Interpretable Human Item Difficulty Prediction	2026-07-11	32 pa... 32 pages, 8 figures, 10 tables
Can AI Reason Like an Urban Planner? Benchmarking Large Language Models Against Professional Judgment	2026-07-11	This ... This paper has been withdrawn by the authors because the current version requires substantial revision and further validation before it can be considered a reliable representation of the work
Beyond Euclidean Clipping: Overcoming Exploration Collapse in LLM RL via Riemannian Isometric Policy Optimization	2026-07-11	ICML 2026
EmoStyle: Affective Conditioning of Style-Specialist Experts for Emotional Image Generation	2026-07-11
Silent Failures in Quantized LLM Reasoning: A Taxonomy-Based Analysis of Hollow Convergence and Failure Mode Shifts	2026-07-10	7 pag... 7 pages, 3 figures, 6 tables
Using LLMs to Adjudicate Static-Analysis Alerts with Error Reduction Techniques	2026-07-10	22 pages
Hierarchical Chain-of-Thought: Enhancing LLM Reasoning Performance and Efficiency	2026-07-10
Practical Source Code Recovery from Binary Functions Using Anchor-Based Retrieval and LLM Reasoning	2026-07-10	12 pages, 5 figures
RepLLM: Toward Automatically Reproducing Network Research Results	2026-07-09	SIGCO... SIGCOMM'26(19 pages, 6 figures, 6 tables)
Hierarchical Control in Multi-Agent Games: LLM-based Planning and RL Execution	2026-07-09	12 pages, 9 figures
TrajAudit: Automated Failure Diagnosis for Agentic Coding Systems	2026-07-09
Can We Trust LLM's Logic? Quantifying Uncertainty, Coherence, and Robustness via a Graph-Based Framework	2026-07-09	42 pa... 42 pages, 14 figures, 12 tables
Fast, Slow, and Tool-augmented Thinking for LLMs: A Review	2026-07-08	The a... The article has been accepted by Frontiers of Computer Science (FCS), with the DOI: {10.1007/s11704-026-51673-0}
MILES: Modular Instruction Memory with Learnable Selection for Self-Improving LLM Reasoning	2026-07-08
Evaluating SageMath-Augmented LLM Agents for Computational and Experimental Mathematics	2026-07-07	37 pa... 37 pages, 16 figures, accepted to 3rd AI for Math Workshop at ICML 2026
Knowing When to Quit: A Principled Framework for Dynamic Abstention in LLM Reasoning	2026-07-07
Open-Ended Scenario Reasoning for Specialist Model Adaptation	2026-07-07
RS-Agent: Automating Remote Sensing Tasks through Intelligent Agent	2026-07-07
AgentsCAD: Automated Design for Manufacturing of FDM Parts via Multi-Agent LLM Reasoning and Geometric Feature Recognition	2026-07-07
ChargeBD: Character-Aware Heterogeneous Agent Reasoning for Guided Engineering in Battery Development	2026-07-06
CAP-CoT: Cycle Adversarial Prompt for Improving Chain of Thoughts in LLM Reasoning	2026-07-06
Think Deep, Not Just Long: Measuring LLM Reasoning Effort via Deep-Thinking Tokens	2026-07-06	Accep... Accepted to ICML 2026
Interactive Learning for LLM Reasoning	2026-07-06	The c... The code is available at https://github.com/linhh29/Interactive-Learning-for-LLM-Reasoning
A Model Can Help Itself: Reward-Free Self-Training for LLM Reasoning	2026-07-05
Scalable Semantic Steering of Embedding Projections	2026-07-04	Accep... Accepted as a short paper at IEEE VIS 2026. 5 pages, 2 figures
rePIRL: Learn PRM with Inverse RL for LLM Reasoning	2026-07-03
LLM-Enhanced Hierarchical Heterogeneous Graph Representation Learning for Malicious Python Package Detection	2026-07-03
R$^2$PO: Decoupling Rollout and Inference Policies for LLM Reasoning	2026-07-03
OpenSIR: Open-Ended Self-Improving Reasoner	2026-07-03

Title	Date	Comment
LinguistAgent Technical Report: A Reflective Multi-Model Platform for Automated Linguistic Annotation	2026-07-21
ResearchArena: Evaluating Sabotage and Monitoring in Automated AI R&D	2026-07-21	51 pages, 11 figures
AIM-CoT: Active Information-driven Multimodal Chain-of-Thought for Vision-Language Reasoning	2026-07-21	Accep... Accepted by ACL 2026 Main Conference. 30 pages, 6 figures
Cognitive Dual-Process Planning for Autonomous Driving with Structured Scene Knowledge and Verifiable Reasoning-Action Consistency	2026-07-21	12 pa... 12 pages, 7 figures. Zhongyao Yang and Haoyu Li contributed equally to this work
DAIS: Dependency-Aware Intermediate QA Supervision for Complex Reasoning	2026-07-21
Measuring Reward-Seeking via Contrastive Belief Updates	2026-07-21	101 p... 101 pages, 66 figures
Not All Errors Are Created Equal: ASCoT Addresses Late-Stage Fragility in Efficient LLM Reasoning	2026-07-21
Guide, Think, Act: Interactive Embodied Reasoning in Vision-Language-Action Models	2026-07-21
CASE: Causal Alignment and Structural Enforcement for Improving Chain-of-Thought Faithfulness	2026-07-21
Bounding Boxes to Improve Small Language Model Performance on Vision-Based Grading Tasks	2026-07-21	Accep... Accepted for 1st Workshop on Small Language Models for Education (SLM4ED '26) at AIED 2026
Rationale-Guided Knowledge Distillation for Cross-Lingual Stance Detection	2026-07-21	23 pa... 23 pages, 7 figures, 3 tables
Robust Reasoning Benchmark	2026-07-21
LatentMT: Machine Translation with Latent Reasoning	2026-07-21
Reasoning Fine-Tuning Induces Persistent Latent Policy States	2026-07-20	Accep... Accepted at the Conference on Language Modeling (COLM) 2026. 45 pages, including appendices; 24 figures and 12 tables. Code: https://github.com/withmartian/mi-cot
Measuring and Mitigating Post-hoc Rationalization in Reverse Chain-of-Thought Generation	2026-07-20	ICML 2026
Value Leakage: An LLM's Answers Are Silently Shaped by Its Own Values	2026-07-20
A Geometric Perspective on Stabilizing Value Conflict Resolution	2026-07-20	Accep... Accepted to ICML Workshop on High-Dimensional Learning Dynamics
Benchmarking Resource-Efficient LLMs for Research Topic Ontology Generation in the Biomedical Field	2026-07-20
SCA: Segment-Wise CoT Compression with Answer Alignment	2026-07-20	9 pag... 9 pages, 5 figures. This version substantially revises the previous preprint with a new method, updated experiments, and rewritten analysis. Code available at the GitHub project repository https://anonymous.4open.science/r/sca-B666
ST-Veto: Spatio-Temporal Token Veto for Diffusion MLLMs via Taylor Prediction and Visual Grounding	2026-07-20	ICML 2026 - main
When a Name Is Not a Name: A Benchmark Dataset and Distilled Reasoning for Culturally Entangled Bangla Homographs in Low-Resource LLMs	2026-07-20
MMR-V: What's Left Unsaid? A Benchmark for Multimodal Deep Reasoning in Videos	2026-07-20	Accep... Accepted at ICLR 2026. Camera-ready version
Reasoning as a Double-Edged Sword: Architecture and Cross-Stage Robustness in Vision-Language-Action Models	2026-07-20
Post-Training Shifts Confidence: A Three-Stage Analysis of How SFT, RL, and OPD Shape CoT Calibration	2026-07-20
Coercion and Deception in AI-to-AI Management: An Agentic Benchmark of Unprompted Escalation	2026-07-20
LenGuard-GPC: Length Guarding with Guided-Prompt Consistency for Spatial Reasoning Reinforce Learning	2026-07-19
Is Your Model Thinking or Just Stagnating? PUMA: Diagnosing Reasoning Pathology via Phase-Momentum Alignment	2026-07-19
RoboPIN: Grounded Embodied Reasoning via Pinned Chain-of-Thought	2026-07-19
Training Continuous Chain of Thought Models: A Tale of Two Regimes	2026-07-18	Accep... Accepted to AdaptFM Workshop, ICML 2026
FUSAR-R1: A Large-Scale Reasoning Model for Intelligent Interpretation of SAR Images	2026-07-18
GRASP: Grounded CoT Reasoning with Dual-Stage Optimization for Multimodal Sarcasm Target Identification	2026-07-18
Logic-Guided Socially-aware Robot Navigation World Model	2026-07-18	The a... The authors have decided to withdraw this manuscript due to concerns regarding its current scope, framing, and presentation. Please do not cite this version
NOWJ@COLIEE 2026: Adaptive Pipelines for Legal Retrieval and Reasoning	2026-07-18	Prese... Presented at COLIEE 2026
Visual Access Boundaries in Vision-Language Model Reasoning	2026-07-18
From Discussion to Execution: Replicating Buggy and Correct Data Science Code	2026-07-18	accep... accepted at the 37th IEEE International Symposium on Software Reliability Engineering (ISSRE 2026)
WeedExpert-R1: Incentivizing Botanical Reasoning in MLLMs with Reinforcement Learning for Precision Weed Grounding	2026-07-17
RIMS: Preference Optimization via Smoothed Multi-pair Aggregation for Small-Scale LLM Retrieval-Augmented Generation	2026-07-17
Audio-Visual Flamingo: Open Audio-Visual Intelligence for Long and Complex Videos	2026-07-17	Proje... Project Page: https://avflamingo.pages.dev/
What Models Express, Suppress, and Resist: Auditing Open-Weight LLMs with Persona Vectors	2026-07-17
Latency-Response Theory Model: Evaluating Large Language Models via Response Accuracy and Chain-of-Thought Length	2026-07-17
MoT: Modularization-of-Thought Prompting for Effective Code Generation	2026-07-17
Length Penalties Make Chain-of-Thought Less Monitorable	2026-07-17
LLM-Guided Transportation Hub Capacity Planning with Textual Business Inputs	2026-07-17
Better Starts, Better Ends: Bootstrapped Iterative Self-Reasoning Distillation for Compressed Reasoning	2026-07-17
Brain-CLIPLM: Semantic Compression for EEG-to-Text Decoding	2026-07-17
Video Streaming Thinking: VideoLLMs Can Watch and Think Simultaneously	2026-07-17	Accep... Accepted by ECCV 2026, project page https://1ranguan.github.io/VST/
ABot-N1: Toward a General Visual Language Navigation Foundation Model	2026-07-17
Neuro-Symbolic AI for LEED compliance: Document-Centric Benchmarking, Deterministic Numeric Checking, and When Multimodal Hurts	2026-07-17
RecGPT-V3 Technical Report	2026-07-17	Technique Report

Title	Date	Comment
Understanding Interpretation Difficulty in Harmful Online Communication: Insights from Cybercrime Communities	2026-07-08
NeuraDock Visual Cognitive Load Agent Tutorial: A Quality-Gated Open-Source EEG Workflow for Alpha Dynamics and Real-Time Applications	2026-06-25	22 pages, 10 figures
Don't Go Breaking My LLM: The Impact of Pruning Attention Layers on Explanation Faithfulness and Confidence Calibration	2026-06-23	Accepted at TMLR
LLM-Based Generalizable Hierarchical Task Planning and Execution for Heterogeneous Robot Teams with Event-Driven Replanning	2026-06-18	full ... full version of this short paper is accepted at Frontiers in Robotics and AI Journal
ICA Lens: Interpreting Language Models Without Training Another Dictionary	2026-06-10	Ongoing Project
Language-Driven Cost Optimization for Autonomous Driving	2026-06-09	Paper... Paper accepted at IEEE Intelligent Transportation Systems Conference (ITSC) 2026
ReflectiChain: Epistemic Grounding in LLM-Driven World Models for Supply Chain Resilience	2026-06-09
MotionGPT-2: A General-Purpose Motion-Language Model for Motion Generation and Understanding	2026-06-08
Analyzing the Correlation Between Hallucinations and Knowledge Conflicts in Large Language Models	2026-06-07
Thinking Through Signs: PEEL as a Semiotic Scaffolding for Epistemically Accountable AI-Enabled Research	2026-06-02	10 pages, 5 figuras
Unveiling the Limits of Large Language Models in Inferring Pragmatic Meaning from Non-Verbal Responses	2026-06-01
The Limits of LLM Forecasting: Parametric Knowledge Gaps Across Conflict Zones	2026-05-29
Hallucination Behavior in Multimodal LLMs Across Agricultural Image Interpretation and Generation Tasks	2026-05-26
How Much Structure Do LLMs Need? Evaluating LLMs for Bibliometric Cluster Description	2026-05-23
Improving Labeling Consistency with Detailed Constitutional Definitions and AI-Driven Evaluation	2026-05-22	Under... Under review at ACL Rolling Review (ARR), May 2026 cycle. Also available at https://doi.org/10.5281/zenodo.20125267
Patch-Effect Graph Kernels for LLM Interpretability	2026-05-07
LLM-ADAM: A Generalizable LLM Agent Framework for Pre-Print Anomaly Detection in Additive Manufacturing	2026-05-05	24 pages, 10 figures
From Research Question to Scientific Workflow: Leveraging Agentic AI for Science Automation	2026-04-23
Agentic AI-Enabled Framework for Thermal Comfort and Building Energy Assessment in Tropical Urban Neighborhoods	2026-04-23	Accep... Accepted at IAQVEC 2026
Learning to Draw ASCII Improves Spatial Reasoning in Language Models	2026-04-16
Arbitration Failure, Not Perceptual Blindness: How Vision-Language Models Resolve Visual-Linguistic Conflicts	2026-04-13
Heterogeneous Graph Importance Scoring and Clustering with Automated LLM-based Interpretation	2026-04-09	26 pa... 26 pages, 11 figures, 8 tables
Runtime Execution Traces Guided Automated Program Repair with Multi-Agent Debate	2026-04-03	12 pa... 12 pages, 4 figures, 8 tables
Discovering Decoupled Functional Modules in Large Language Models	2026-03-18	AAAI-26 Oral
LaMoGen: Language to Motion Generation Through LLM-Guided Symbolic Inference	2026-03-12	Accep... Accepted by CVPR 2026. Supplementary material included. Project page: https://jjkislele.github.io/LaMoGen/
Pneuma-Seeker: A Relational Reification Mechanism to Align AI Agents with Human Work over Relational Data	2026-03-11
PhysLLM: Harnessing Large Language Models for Cross-Modal Remote Physiological Sensing	2026-03-05	Accep... Accepted by International Conference on Learning Representations (ICLR) 2026
Cache What Lasts: Token Retention for Memory-Bounded KV Cache in LLMs	2026-03-01
Sparse Shift Autoencoders for Identifying Concepts from Large Language Model Activations	2026-02-27	27 pages, 9 figures
Co-Disclosing the Computer: LLM-Mediated Computing through Reflective Conversation	2026-02-27	CHI'26
Neuro-Symbolic Control with Large Language Models for Language-Guided Spatial Tasks	2026-02-21
Stroke Lesions as a Rosetta Stone for Language Model Interpretability	2026-02-03	45 pages, 17 figures
Uncertainty-Aware Gradient Signal-to-Noise Data Selection for Instruction Tuning	2026-01-20	Preprint
A Shared Geometry of Difficulty in Multilingual Language Models	2026-01-19
Causal-SAM-LLM: Large Language Models as Causal Reasoners for Robust Medical Segmentation	2026-01-16	Accep... Accepted by IEEE ICASSP 2026
Self-reflection in Automated Qualitative Coding: Improving Text Annotation through Secondary LLM Critique	2026-01-14
Word Synchronization Challenge: A Benchmark for Word Association Responses for Large Language Models	2026-01-14
Can LLMs interpret figurative language as humans do?: surface-level vs representational similarity	2026-01-14	17 pages, 5 figures
Neuro-Symbolic Compliance: Integrating LLMs and SMT Solvers for Automated Financial Legal Analysis	2026-01-07	10 pa... 10 pages, 6 tables, 3 figures, accepted by the 2nd ACM AIware Conference
LLM Interpretability with Identifiable Temporal-Instantaneous Representation	2026-01-02	NeurIPS 2025
Trust Me, I Know This Function: Hijacking LLM Static Analysis using Bias	2025-12-18
Verification-Guided Context Optimization for Tool Calling via Hierarchical LLMs-as-Editors	2025-12-15	Accep... Accepted by AAAI 2026 Workshop on Agentic AI Benchmarks and Applications for Enterprise Tasks
Visualizing token importance for black-box language models	2025-12-12
SpeechIQ: Speech-Agentic Intelligence Quotient Across Cognitive Levels in Voice Understanding by Large Language Models	2025-12-01	ACL 2... ACL 2025 main. Our Speech-IQ leaderboard is hosted at huggingface.co/spaces/nvidia/Speech-IQ-leaderboard. Speech-IQ Calculator: https://github.com/YukinoWan/SpeechIQ
For Those Who May Find Themselves on the Red Team	2025-11-23
HSKBenchmark: Modeling and Benchmarking Chinese Second Language Acquisition in Large Language Models through Curriculum Tuning	2025-11-19	Accep... Accepted by AAAI-2026
KnowThyself: An Agentic Assistant for LLM Interpretability	2025-11-05	5 pag... 5 pages, 1 figure, Accepted for publication at the Demonstration Track of the 40th AAAI Conference on Artificial Intelligence (AAAI 26)
Imperfect Language, Artificial Intelligence, and the Human Mind: An Interdisciplinary Approach to Linguistic Errors in Native Spanish Speakers	2025-11-03	12 pages, 3 figures
To Sink or Not to Sink: Visual Information Pathways in Large Vision-Language Models	2025-10-09	Prepr... Preprint. Project page: https://davidhalladay.github.io/diysink_demo
Automated Repair of Ambiguous Problem Descriptions for LLM-Based Code Generation	2025-09-24

Title	Date	Comment
LLM-Grounded Explainable AI for Supply Chain Risk Early Warning via Temporal Graph Attention Networks	2026-07-21
Automated Data Engineering and Feature Selection for the Case Study of Warpage Detection in Fused Deposition Modeling	2026-07-20
Explaining and Tuning Transformer-based LLMs in Arithmetic Tasks with Human Strategies	2026-07-19
FST.ai 2.5: Explainable and Uncertainty-Aware AI for Olympic and Para-Taekwondo Decision Support, Athlete Digital Twins, and Federation-Scale Analytics	2026-07-18	14 pages, 4 figures
FAIR_XAI: Improving Multimodal Foundation Model Fairness via Explainability for Wellbeing Assessment	2026-07-17	11 pages, 4 figures
VeriX-Anon: A Multi-Layered Framework for Mathematically Verifiable Outsourced Target-Driven Data Anonymization	2026-07-17	v2: r... v2: revised after peer review. Evaluation expanded from 3 to 7 datasets, per-dataset Wasserstein-threshold calibration added, effect-size CIs and cross-dataset statistics reported, and analytical zk-SNARK/MPC/TEE baselines added. Minor errors corrected
Scaling Time Series Classification via XAI-Driven Data Reduction	2026-07-17	Accep... Accepted for AALTD workshop at ECML-PKDD 2026
Automated identification of Ichneumonoidea wasps via YOLO-based deep learning: Integrating HiresCam for Explainable AI	2026-07-16	15 pages, 20 figures
Explaining Process Control Optimisation Recommendations via GradientSHAP and Implicit Differentiation	2026-07-16
Diagnosing and Mitigating Domain Shift in Permission-Based Android Malware Detection	2026-07-15
"Trust Junk" Leads to Unjustified Support for Highly Discriminatory Predictive Models	2026-07-14
MLPTR-CC: Multi-label Pathology Test Recommendation using Classifier Chains and SHAP	2026-07-14
Atomic Units of X: The Compression Layer of Intelligence	2026-07-14
Evaluating RE Practices for Explainability: Synthesizing Insights from Daimler Truck into an Explainable RE Framework Proposal	2026-07-13
Detecting Explanatory Insufficiency in Learned Representations: A Framework for Representational Vigilance	2026-07-13	25 pa... 25 pages, 2 figures, no tables, 16 references. Conceptual and methodological framework for monitoring representational adequacy and detecting explanatory insufficiency in learned representations
ConceptSMILE: Auditing the Trustworthiness of Concept-Based Explainable AI	2026-07-10
Knowledge Graphs and Explainable AI as Complementary Resources for Urban Mining	2026-07-10	Accep... Accepted for presentation at the AISE Workshop @ IJCAI-ECAI 2026
All Explanations are Wrong, But Many Are Useful: Exploring the Rashomon Explanation Set with Large Language Models	2026-07-10
Application of machine learning to monster level prediction in tabletop RPG game design	2026-07-10
Unlearning to Protect: A Distilled Reinforcement Learning Framework with Privacy-Preserving Feature Unlearning and XAI for IoT Security	2026-07-09
The Contribution of XAI for the Safe Development and Certification of AI: An Expert-Based Analysis	2026-07-09	This ... This preprint has not undergone peer review or any post-submission improvements or corrections. The Version of Record of this contribution is published in Artificial Intelligence in HCI (HCII 2026), Lecture Notes in Computer Science, vol. 16745, and is available online at https://doi.org/10.1007/978-3-032-30849-8_13
Mechanistic Interpretability for Neural Networks: Circuits, Sparse Features and Symbolic Reasoning	2026-07-08	20 pages
Why Fake ? Unveiling the Semantic Vocabulary of Deepfake Detectors	2026-07-08	Accep... Accepted at CVPRW 2026
Large Language Models (LLMs) and Generative AI in Cybersecurity and Privacy: A Survey of Dual-Use Risks, AI-Generated Malware, Explainability, and Defensive Strategies	2026-07-08	Invit... Invited survey paper. 10 pages, 5 figures, 2 tables
Reduced NEXI protocol for the quantification of human gray matter microstructure on the Connectome 2.0 scanner	2026-07-07	Submi... Submitted to Imaging Neuroscience. This all-in-one version includes supplementary materials. 34 pages, 145 figures, 4 tables
Explainable embeddings with Distance Explainer	2026-07-07	21 pa... 21 pages, 12 figures. Accepted to the 4th World Conference on eXplainable Artificial Intelligence. Method implementation: https://research-software-directory.org/software/distance-explainer
Measuring What Matters: A Unified Evaluation Framework for GNN Explainability	2026-07-06
Dynamic Interest Rate Discovery in Decentralized Finance: A Reverse Kelly Automated Market Maker for Risk-Adjusted Lending	2026-07-05
Explainable AI for Screening Abuse-Related Trauma in Bangladeshi Children: A Training-Free Multimodal Framework Evaluated on Noise-Aware Synthetic Data	2026-07-04	6 pages, 5 figures
Activation-Deactivation: A General Framework for Robust Post-hoc Explainable AI	2026-07-04	Prepr... Preprint: Under Review; Updated experiments & Figures
Better Together? The Role of Explanations in Supporting Novices in Individual and Collective Deliberations about AI	2026-07-03	30 pa... 30 pages main text, 8 figures, 4 tables. Supplementary material is included in the appendix
Algebraic Model Counting for Global Analysis of Optimal Decision Trees	2026-07-02	Proc.... Proc. Joint European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML-PKDD 2026), LNCS, Naples, Italy, 7-11 September 2026
Quantum-Inspired Vision: Leveraging Wave-Particle Duality for Low-Illumination Enhancement	2026-07-02
Reducing Labeling Effort in Architecture Technical Debt Detection through Active Learning and Explainable AI	2026-07-01	Accep... Accepted for publication in Empirical Software Engineering (EMSE) Journal, 2026
PACE: A Neuro-Symbolic Framework for Plausible and Actionable Counterfactual Explanations	2026-07-01
CPG-PAD: Concept-Informed Prompts Guided Presentation Attack Detection	2026-07-01	Accep... Accepted by IEEE Transactions on Information Forensics & Security (TIFS)
Understanding Large Language Models	2026-07-01	25 pages, 1 figure
Explainable AI for Cancer Drug Response Prediction: Beyond Univariate Feature Attributions	2026-07-01
FedXDS: Leveraging Model Attribution Methods to counteract Data Heterogeneity in Federated Learning	2026-06-30
Scientific Explanations in Health Sciences: Causality, Trust, and Epistemic Adequacy	2026-06-30
Explainable AI in Speaker Recognition -- Making Latent Representations Understandable	2026-06-29	A working paper
Feature-level Interaction Explanations in Multimodal Transformers	2026-06-28
Understanding LLM Intervention Explanations in Multi-Party Human-Robot Interaction	2026-06-28	Accep... Accepted for 2026 36th IEEE International Conference on Robot and Human Interactive Communication
Low-cost concept-based localized explanations: How far can we get with training-free approaches?	2026-06-27	6 pag... 6 pages, 2 figures, 4 tables. Accepted at the 2026 IEEE International Conference on Artificial Intelligence (CAI), 8-10 May 2026, Granada, Spain. Code: https://github.com/darianfgUgr/CoNa

Title	Date	Comment
CircuitKIT : Circuit Discovery, Evaluation, and Application Toolkit for Mechanistic Interpretability	2026-07-21
CLT-Forge: A Scalable Library for Cross-Layer Transcoders and Attribution Graphs	2026-07-21	9 pag... 9 pages, 7 figures, 1 table. Code: https://github.com/LLM-Interp/CLT-Forge. Demonstration video: https://youtu.be/6ptrrLawTl8
EvalSafetyGap: A Hybrid Survey and Conceptual Framework for LLM Evaluation-Safety Failures	2026-07-21	This ... This manuscript is a 80-page hybrid survey and conceptual framework on LLM evaluation and AI-safety failures. It includes 8 figures and multiple evidence-synthesis tables, covering literature from 2018 to 2026. The paper introduces the EvalSafetyGap framework and reports a structured audit of 10 LLMs. It is submitted as a review/survey article and is not currently under consideration elsewhere
Breaking the Block: Preserving Data Continuity to Train Superior SAEs for Instruct Models	2026-07-20
Measuring Monosemanticity in Sparse Autoencoders via Latent Activation Coherence	2026-07-20	This ... This is a preprint version. A shorter version of this paper has been accepted for presentation and publication in the post-workshop proceedings of the 8th International Workshop on eXplainable Knowledge Discovery in Data Mining (XKDD 2026), co-located with ECML PKDD 2026. The appendix is included only in this preprint and is not part of the peer-reviewed proceedings paper
Every Component is a Lookup: Token Attribution and Composition from a Single Decomposition	2026-07-19
What Do They See? Interpreting Complex Road Scenarios Through the Eyes of Vision-Language-Action Models for Safe and Trustworthy Autonomous Vehicle Learning	2026-07-18
Are Arithmetic Heuristic Neurons Form-Invariant? A Mechanistic Analysis of Symbols, Text, and Code in LLMs	2026-07-18	Under Review
Mechanistic Interpretability of Cognitive Complexity in LLMs via Linear Probing using Bloom's Taxonomy	2026-07-17	Prepr... Preprint. Under review
Analysing Moral Bias in Finetuned LLMs through Mechanistic Interpretability	2026-07-17	Prepr... Preprint. Under review
Hide and Seek in Embedding Space: Geometry-based Steganography and Detection in Large Language Models	2026-07-17
Steering Robustness into World Action Models via Mechanistic Interpretability and Optimal Control	2026-07-16
Transcoders for Investigating Deception in Language Models	2026-07-16
Tracing LLM Behavior to the Training Data with Empirical Next-Token Distributions	2026-07-15
From Observation to Insight: Mechanistic World Models and the Quest for Autonomous Discovery	2026-07-15
Hierarchical Latent Structures in Data Generation Process Unify Mechanistic Phenomena across Scale	2026-07-14
Same Compression Principle, Different Geometry: Rate-Distortion Signatures Dissociate Biological and Artificial Visual Systems	2026-07-13
Inside the Unfair Judge: A Mechanistic Interpretability Account of LLM-as-Judge Bias	2026-07-13	58 pa... 58 pages, 13 figures, 30 tables; project page: https://xzx34.github.io/unfair-judge/
Emergent Hierarchical Monosemantic Neurons from the Group-Contrastive Forward-Forward Algorithm	2026-07-13
Which Neurons Detect Malicious Code? A Probing Study of LLM Security Knowledge	2026-07-11	The p... The paper has been peer reviewed and accepted for publication in the 20th International Symposium on Empirical Software Engineering and Measurement (ESEM 2026)
MLPs are Hebbians: Constructing Efficient Fact-Storing MLPs for Transformers	2026-07-10
XAI and Statistical Analysis for Reliable Intrusion Detection in the UAVIDS-2025 Dataset: From Tree to Hybrid and Tabular DNN Ensembles	2026-07-10	Accep... Accepted at IEEE CITS 2026, Greece
When Structured Sparse Autoencoders Learn Consistent Concepts Across Modalities	2026-07-09
Cross-seed explainability using Procrustes-conditioned Joint End-to-end Top-K Sparse Autoencoders	2026-07-09	17 pa... 17 pages, 4 figures, 6 tables
Towards Isolated Interventions via Almost Orthogonal Features in Language Models	2026-07-09	Accep... Accepted as a conference paper at the Conference on Language Modeling (COLM) 2026
Certified Interventional Fidelity: Anytime-Valid, Adaptive Evaluation of Causal Claims in Mechanistic Interpretability	2026-07-09	Accep... Accepted at UAI 2026 (Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence). Code: https://github.com/AsiaeeLab/certified-interventional-fidelity
Diagnosing Shape-Prior Shortcuts in Long-Range Single-Shot Fringe Projection Profilometry	2026-07-09	21 pages, 13 figures
Mechanistic Interpretability of LLM Jailbreaks via Internal Attribution Graphs	2026-07-08
Temporal Preference Concepts and their Functions in a Large Language Model	2026-07-08
How Learning Dynamics Drive Adversarially Robust Generalization?	2026-07-08	Accep... Accepted at the 42nd Conference on Uncertainty in Artificial Intelligence (UAI 2026)
Mechanistic Interpretability for Neural Networks: Circuits, Sparse Features and Symbolic Reasoning	2026-07-08	20 pages
Latent Programming Horizons in Coding Agents	2026-07-06
Beyond the Black Box: Interpretability of Agentic AI Tool Use	2026-07-05	12 pa... 12 pages, 4 figures, 17 tables
Interpretability and Generalization Bounds for Learning Spatial Physics	2026-07-05	To ap... To appear in ICML 2026. 18 pages, 13 figures
Cultural Binding Heads in Language Models	2026-07-04
Using Mechanistic Interpretability to Craft Adversarial Attacks against Large Language Models	2026-07-03
Individual Parameters in Weight-Sparse Transformers Appear Interpretable	2026-07-03	20 pa... 20 pages, 19 figures, 3 tables. Project website: https://weightpedia.org/individual-parameters-in-sparse-transformers/
MatPhaseBench: A Semantics-Guided Benchmark for Materials Phase Diagrams Understanding	2026-07-03
Induction Heads Interpolate N-Grams	2026-07-02	Publi... Published as a conference paper at ICML 2026. OpenReview: https://openreview.net/forum?id=BSY7jhBxM1
Towards Robustness against Typographic Attack with Training-free Concept Localization	2026-07-02	15 pa... 15 pages main text, provisionally accepted to ECCV 2026
Conditional Co-Ablation: Recovering Self-Repair Backups in Transformer Circuits	2026-07-02
Shared Semantics, Divergent Mechanisms: Unsupervised Feature Discovery by Aligning Semantics and Mechanisms	2026-07-02	40 pa... 40 pages; accepted as an ICML 2026 Spotlight; project page: https://merenova.github.io/distribution-level-feature-discovery/
Expander Sparse Autoencoders: Parameter-Efficient Dictionaries for Mechanistic Interpretability	2026-07-02
Mechanistic Interpretability and Causal Feature Steering of Neural Quantum States via Sparse Autoencoders	2026-07-01	15 pa... 15 pages, 7 figures. Comments welcome!
Muon as a Residual Connection	2026-07-01
Interpreting Global Perturbation Robustness of Image Models using Axiomatic Spectral Importance Decomposition	2026-07-01	Accep... Accepted by Transactions on Machine Learning Research (TMLR 2024)
MetaOthello: A Controlled Study of Multiple World Models in Transformers	2026-07-01	Camer... Camera-ready version. Accepted to the 43rd International Conference on Machine Learning (ICML 2026)
Representation as a Bottleneck for Mechanistic Interpretability: The Manifestation Unit Protocol	2026-06-30	65 pa... 65 pages. Interactive demos: https://manifestation-xai.github.io/manifestation-transformers/ , https://manifestation-xai.github.io/manifestation-cnn
Surrogate Fidelity: When Can Open LLMs Explain Closed Ones?	2026-06-30

Navigation Menu

Latest 50 Papers - July 23, 2026 #124

LLM Reasoning

Chain of Thought

LLM Interpretability

Explainable AI

Mechanistic Interpretability

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Latest 50 Papers - July 23, 2026 #124

Description

LLM Reasoning

Chain of Thought

LLM Interpretability

Explainable AI

Mechanistic Interpretability

Metadata

Metadata

Assignees

Labels

Projects

Milestone

Relationships

Development

Issue actions