{"total":77,"rows":[{"id":"2403.19647","title":"Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models","created":"2024-03-28","cited_by_count":435,"area":null},{"id":"2404.14082","title":"Mechanistic Interpretability for AI Safety -- A Review","created":"2024-04-22","cited_by_count":422,"area":"mechinterp"},{"id":"2405.14860","title":"Not All Language Model Features Are One-Dimensionally Linear","created":"2024-05-23","cited_by_count":212,"area":null},{"id":"2406.11944","title":"Transcoders Find Interpretable LLM Feature Circuits","created":"2024-06-17","cited_by_count":200,"area":"mechinterp"},{"id":"2406.19384","title":"The Remarkable Robustness of LLMs: Stages of Inference?","created":"2024-06-27","cited_by_count":146,"area":null},{"id":"2407.02646","title":"A Practical Review of Mechanistic Interpretability for Transformer-Based Language Models","created":"2024-07-02","cited_by_count":118,"area":"mechinterp"},{"id":"2410.06205","title":"Round and Round We Go! What makes Rotary Positional Encodings useful?","created":"2024-10-08","cited_by_count":115,"area":null},{"id":"2401.12181","title":"Universal Neurons in GPT2 Language Models","created":"2024-01-22","cited_by_count":102,"area":"mechinterp"},{"id":"2410.21272","title":"Arithmetic Without Algorithms: Language Models Solve Math With a Bag of Heuristics","created":"2024-10-28","cited_by_count":97,"area":"lr"},{"id":"2411.00348","title":"Attention Tracker: Detecting Prompt Injection Attacks in LLMs","created":"2024-11-01","cited_by_count":95,"area":"safety"},{"id":"2405.08366","title":"Towards Principled Evaluations of Sparse Autoencoders for Interpretability and Control","created":"2024-05-14","cited_by_count":86,"area":"mechinterp"},{"id":"2405.00208","title":"A Primer on the Inner Workings of Transformer-based Language Models","created":"2024-04-30","cited_by_count":85,"area":null},{"id":"2410.13708","title":"On the Role of Attention Heads in Large Language Model Safety","created":"2024-10-17","cited_by_count":72,"area":"safety"},{"id":"2409.03752","title":"Attention Heads of Large Language Models: A Survey","created":"2024-09-05","cited_by_count":67,"area":null},{"id":"2407.10827","title":"LLM Circuit Analyses Are Consistent Across Training and Scale","created":"2024-07-15","cited_by_count":63,"area":null},{"id":"2407.15017","title":"Knowledge Mechanisms in Large Language Models: A Survey and Perspective","created":"2024-07-22","cited_by_count":51,"area":null},{"id":"2505.13898","title":"Do Language Models Use Their Depth Efficiently?","created":"2025-05-20","cited_by_count":50,"area":null},{"id":"2410.01104","title":"Softmax is not Enough (for Sharp Size Generalisation)","created":"2024-10-01","cited_by_count":46,"area":null},{"id":"2410.02984","title":"Differentiation and Specialization of Attention Heads via the Refined Local Learning Coefficient","created":"2024-10-03","cited_by_count":40,"area":null},{"id":"2312.12141","title":"Neuron-Level Knowledge Attribution in Large Language Models","created":"2023-12-19","cited_by_count":37,"area":null},{"id":"2503.09543","title":"PolyPythias: Stability and Outliers across Fifty Language Model Pre-Training Runs","created":"2025-03-12","cited_by_count":35,"area":null},{"id":"2410.11781","title":"Language Models Encode Numbers Using Digit Representations in Base 10","created":"2024-10-15","cited_by_count":32,"area":null},{"id":"2409.14144","title":"Interpreting Arithmetic Mechanism in Large Language Models through Comparative Neuron Analysis","created":"2024-09-21","cited_by_count":31,"area":null},{"id":"2406.16320","title":"What Do VLMs NOTICE? A Mechanistic Interpretability Pipeline for Gaussian-Noise-free Text-Image Corruption and Evaluation","created":"2024-06-24","cited_by_count":30,"area":"mllm"},{"id":"2405.16229","title":"No Two Devils Alike: Unveiling Distinct Mechanisms of Fine-tuning Attacks","created":"2024-05-25","cited_by_count":29,"area":"alignment"},{"id":"2408.09503","title":"Out-of-distribution generalization via composition: a lens through induction heads in Transformers","created":"2024-08-18","cited_by_count":28,"area":null},{"id":"2407.04690","title":"Missed Causes and Ambiguous Effects: Counterfactuals Pose Challenges for Interpreting Neural Networks","created":"2024-07-05","cited_by_count":23,"area":null},{"id":"2502.10835","title":"Back Attention: Understanding and Enhancing Multi-Hop Reasoning in Large Language Models","created":"2025-02-15","cited_by_count":23,"area":"lr"},{"id":"2502.11812","title":"Towards Understanding Fine-Tuning Mechanisms of LLMs via Circuit Analysis","created":"2025-02-17","cited_by_count":20,"area":null},{"id":"2504.10063","title":"Hallucination Detection in LLMs with Topological Divergence on Attention Graphs","created":"2025-04-14","cited_by_count":19,"area":null},{"id":"2412.11965","title":"Inferring Functionality of Attention Heads from their Parameters","created":"2024-12-16","cited_by_count":17,"area":null},{"id":"2502.19981","title":"The Lookahead Limitation: Why Multi-Operand Addition is Hard for LLMs","created":"2025-02-27","cited_by_count":15,"area":null},{"id":"2501.14457","title":"Understanding and Mitigating Gender Bias in LLMs via Interpretable Neuron Editing","created":"2025-01-24","cited_by_count":14,"area":null},{"id":"2406.03345","title":"Feature contamination: Neural networks learn uncorrelated features and fail to generalize","created":"2024-06-05","cited_by_count":11,"area":null},{"id":"2411.10950","title":"Understanding Multimodal LLMs: the Mechanistic Interpretability of Llava in Visual Question Answering","created":"2024-11-17","cited_by_count":11,"area":"mllm"},{"id":"2502.05911","title":"GRAIT: Gradient-Driven Refusal-Aware Instruction Tuning for Effective Hallucination Mitigation","created":"2025-02-09","cited_by_count":11,"area":"safety"},{"id":"2503.11302","title":"Are formal and functional linguistic mechanisms dissociated in language models?","created":"2025-03-14","cited_by_count":11,"area":null},{"id":"2502.11244","title":"Soteria: Language-Specific Functional Parameter Steering for Multilingual Safety Alignment","created":"2025-02-16","cited_by_count":10,"area":"alignment"},{"id":"2602.20708","title":"ICON: Indirect Prompt Injection Defense for Agents based on Inference-Time Correction","created":"2026-02-24","cited_by_count":10,"area":"agents"},{"id":"2504.20938","title":"Towards Understanding the Nature of Attention with Low-Rank Sparse Decomposition","created":"2025-04-29","cited_by_count":9,"area":"mechinterp"},{"id":"2505.05145","title":"Understanding In-context Learning of Addition via Activation Subspaces","created":"2025-05-08","cited_by_count":8,"area":null},{"id":"2405.17420","title":"Survival of the Fittest Representation: A Case Study with Modular Addition","created":"2024-05-27","cited_by_count":7,"area":null},{"id":"2407.11542","title":"Counting in Small Transformers: The Delicate Interplay between Attention and Feed-Forward Layers","created":"2024-07-16","cited_by_count":6,"area":null},{"id":"2502.13946","title":"Why Safeguarded Ships Run Aground? Aligned Large Language Models' Safety Mechanisms Tend to Be Anchored in The Template Region","created":"2025-02-19","cited_by_count":6,"area":"alignment"},{"id":"2506.02546","title":"To trust or not to trust: Attention-based Trust Management for LLM Multi-Agent Systems","created":"2025-06-03","cited_by_count":6,"area":"agents"},{"id":"2508.02513","title":"Modular Arithmetic: Language Models Solve Math Digit by Digit","created":"2025-08-04","cited_by_count":6,"area":null},{"id":"2508.16929","title":"Dimensional Collapse in Transformer Attention Outputs: A Challenge for Sparse Dictionary Learning","created":"2025-08-23","cited_by_count":6,"area":"mechinterp"},{"id":"2601.10266","title":"Measuring Affinity between Attention-Head Weight Subspaces via the Projection Kernel","created":"2026-01-15","cited_by_count":5,"area":null},{"id":"2506.17052","title":"From Concepts to Components: Concept-Agnostic Attention Module Discovery in Transformers","created":"2025-06-20","cited_by_count":4,"area":null},{"id":"2502.14171","title":"Enhancing Conversational Agents with Theory of Mind: Aligning Beliefs, Desires, and Intentions for Human-Like Interaction","created":"2025-02-20","cited_by_count":3,"area":"agents"},{"id":"2508.20040","title":"Model Science: getting serious about verification, explanation and control of AI systems","created":"2025-08-27","cited_by_count":3,"area":null},{"id":"2605.24059","title":"Spectral Probe-Circuits: A Three-Step Recipe for Identifying Attention-Head Circuits in Pretrained Transformers","created":"2026-05-22","cited_by_count":3,"area":"mechinterp"},{"id":"2510.17256","title":"Explainability of Large Language Models: Opportunities and Challenges toward Generating Trustworthy Explanations","created":"2025-10-20","cited_by_count":2,"area":"mechinterp"},{"id":"2603.10771","title":"Word Recovery in Large Language Models Enables Character-Level Tokenization Robustness","created":"2026-03-11","cited_by_count":2,"area":null},{"id":"2606.05378","title":"Pattern Selectivity is Not Task-Causal Structure: A Cross-Architecture Mechanistic Study of Composed-Task Circuits in 1B-Class Language Models","created":"2026-06-03","cited_by_count":2,"area":"moe"},{"id":"2411.07071","title":"Universal Response and Emergence of Induction in LLMs","created":"2024-11-11","cited_by_count":1,"area":null},{"id":"2411.15281","title":"ElastiFormer: Learned Redundancy Reduction in Transformer via Self-Distillation","created":"2024-11-22","cited_by_count":1,"area":"lr"},{"id":"2510.10489","title":"Head-wise Adaptive Rotary Positional Encoding for Fine-Grained Image Generation","created":"2025-10-12","cited_by_count":1,"area":null},{"id":"2512.10547","title":"Unlocking the Address Book: Dissecting the Sparse Semantic Structure of LLM Key-Value Caches via Sparse Autoencoders","created":"2025-12-11","cited_by_count":1,"area":"inference"},{"id":"2602.07778","title":"Attn-GS: Attention-Guided Context Compression for Efficient Personalized LLMs","created":"2026-02-08","cited_by_count":1,"area":"inference"},{"id":"2607.16693","title":"Are Arithmetic Heuristic Neurons Form-Invariant? A Mechanistic Analysis of Symbols, Text, and Code in LLMs","created":"2026-07-18","cited_by_count":1,"area":"mechinterp"},{"id":"2502.19669","title":"Investigating Neurons and Heads in Transformer-based LLMs for Typographical Errors","created":"2025-02-27","cited_by_count":0,"area":null},{"id":"2502.20499","title":"Data Distributional Properties As Inductive Bias for Systematic Generalization","created":"2025-02-27","cited_by_count":0,"area":null},{"id":"2505.10822","title":"Distilled Circuits: A Mechanistic Study of Internal Restructuring in Knowledge Distillation","created":"2025-05-16","cited_by_count":0,"area":"distill"},{"id":"2508.06352","title":"Leveraging Generative AI for Human Understanding: Meta-Requirements and Design Principles for Explanatory AI as a new Paradigm","created":"2025-08-08","cited_by_count":0,"area":null},{"id":"2510.19427","title":"Revisiting the Relation Between Robustness and Universality","created":"2025-10-22","cited_by_count":0,"area":null},{"id":"2602.08548","title":"How Do Language Models Understand Tables? A Mechanistic Analysis of Cell Location","created":"2026-02-09","cited_by_count":0,"area":null},{"id":"2602.17526","title":"The Anxiety of Influence: Bloom Filters in Transformer Attention Heads","created":"2026-02-19","cited_by_count":0,"area":null},{"id":"2602.22453","title":"Identifying Crucial Attention Heads for Multilingual Language Models: Retrieval and Retrieval-Transition Heads","created":"2026-02-25","cited_by_count":0,"area":null},{"id":"2604.03764","title":"Automated Attention Pattern Discovery at Scale in Large Language Models","created":"2026-04-04","cited_by_count":0,"area":null},{"id":"2605.28825","title":"MechELK: A Mechanistic Interpretability Framework for Eliciting Latent Knowledge in Large Language Models","created":"2026-04-07","cited_by_count":0,"area":"lr"},{"id":"2608.15809","title":"A Pre-Specified Construction-Confirmation Test of Operation-Level Causal Transfer Across Finite Isomorphic Symbolic Domains","created":"2026-08-16","cited_by_count":0,"area":null},{"id":"2609.01170","title":"Pre-carved Niches: The Formation Dynamics of Modular Task Partitions in Early LLM Training","created":"2026-09-01","cited_by_count":0,"area":null},{"id":"2609.09902","title":"Contrastive Projection: Reading Transformer Internals by Differencing Logit Lenses","created":"2026-09-09","cited_by_count":0,"area":"mechinterp"},{"id":"2609.32964","title":"The Commit-Abstain Circuit: Why Language Models Hallucinate Instead of Abstaining","created":"2026-09-26","cited_by_count":0,"area":"mechinterp"},{"id":"2609.33150","title":"Generalization Dynamics of LM Pre-training","created":"2026-09-27","cited_by_count":0,"area":null},{"id":"2609.33851","title":"Rethinking Contextualization by Reinterpreting Attention Head Channels","created":"2026-09-27","cited_by_count":0,"area":"lr"}],"next_cursor":null}