{"list":"other","total":91,"rows":[{"id":"1705.07874","title":"A Unified Approach to Interpreting Model Predictions","created":"2017-05-22","cited_by_count":6555},{"id":"1602.04938","title":"\"Why Should I Trust You?\": Explaining the Predictions of Any Classifier","created":"2016-02-16","cited_by_count":5070},{"id":"1703.01365","title":"Axiomatic Attribution for Deep Networks","created":"2017-03-04","cited_by_count":3384},{"id":"2307.15043","title":"Universal and Transferable Adversarial Attacks on Aligned Language Models","created":"2023-07-27","cited_by_count":3104},{"id":"2202.05262","title":"Locating and Editing Factual Associations in GPT","created":"2022-02-10","cited_by_count":2691},{"id":"1312.6034","title":"Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps","created":"2013-12-20","cited_by_count":2371},{"id":"2309.08600","title":"Sparse Autoencoders Find Highly Interpretable Features in Language Models","created":"2023-09-15","cited_by_count":1420},{"id":"1702.08608","title":"Towards A Rigorous Science of Interpretable Machine Learning","created":"2017-02-28","cited_by_count":1416},{"id":"1610.01644","title":"Understanding intermediate layers using linear classifier probes","created":"2016-10-05","cited_by_count":1353},{"id":"1711.11279","title":"Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV)","created":"2017-11-30","cited_by_count":1319},{"id":"1706.07269","title":"Explanation in Artificial Intelligence: Insights from the Social Sciences","created":"2017-06-22","cited_by_count":1206},{"id":"2211.00593","title":"Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small","created":"2022-11-01","cited_by_count":1165},{"id":"1810.03292","title":"Sanity Checks for Saliency Maps","created":"2018-10-08","cited_by_count":1156},{"id":"2212.04089","title":"Editing Models with Task Arithmetic","created":"2022-12-08","cited_by_count":1120},{"id":"2005.04118","title":"Beyond Accuracy: Behavioral Testing of NLP models with CheckList","created":"2020-05-08","cited_by_count":1006},{"id":"2210.07229","title":"Mass-Editing Memory in a Transformer","created":"2022-10-13","cited_by_count":1001},{"id":"2209.10652","title":"Toy Models of Superposition","created":"2022-09-21","cited_by_count":971},{"id":"1811.12359","title":"Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations","created":"2018-11-29","cited_by_count":941},{"id":"2102.12452","title":"Probing Classifiers: Promises, Shortcomings, and Advances","created":"2021-02-24","cited_by_count":876},{"id":"2304.14997","title":"Towards Automated Circuit Discovery for Mechanistic Interpretability","created":"2023-04-28","cited_by_count":724},{"id":"1606.03490","title":"The Mythos of Model Interpretability","created":"2016-06-10","cited_by_count":721},{"id":"2311.03658","title":"The Linear Representation Hypothesis and the Geometry of Large Language Models","created":"2023-11-07","cited_by_count":716},{"id":"2310.06824","title":"The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets","created":"2023-10-10","cited_by_count":712},{"id":"1909.03368","title":"Designing and Interpreting Probes with Control Tasks","created":"2019-09-08","cited_by_count":640},{"id":"2004.03685","title":"Towards Faithfully Interpretable NLP Systems: How should we define and evaluate faithfulness?","created":"2020-04-07","cited_by_count":599},{"id":"1907.04809","title":"Variational Autoencoders and Nonlinear ICA: A Unifying Framework","created":"2019-07-10","cited_by_count":503},{"id":"1805.01070","title":"What you can cram into a single vector: Probing sentence embeddings for linguistic properties","created":"2018-05-03","cited_by_count":457},{"id":"2403.19647","title":"Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models","created":"2024-03-28","cited_by_count":435},{"id":"2106.02997","title":"Causal Abstractions of Neural Networks","created":"2021-06-06","cited_by_count":422},{"id":"2404.14082","title":"Mechanistic Interpretability for AI Safety -- A Review","created":"2024-04-22","cited_by_count":422},{"id":"2309.16042","title":"Towards Best Practices of Activation Patching in Language Models: Metrics and Methods","created":"2023-09-27","cited_by_count":359},{"id":"1711.00867","title":"The (Un)reliability of saliency methods","created":"2017-11-02","cited_by_count":333},{"id":"2106.04619","title":"Self-Supervised Learning with Data Augmentations Provably Isolates Content from Style","created":"2021-06-08","cited_by_count":312},{"id":"2310.15213","title":"Function Vectors in Large Language Models","created":"2023-10-23","cited_by_count":308},{"id":"1605.01713","title":"Not Just a Black Box: Learning Important Features Through Propagating Activation Differences","created":"2016-05-05","cited_by_count":297},{"id":"1301.2300","title":"Direct and Indirect Effects","created":"2013-01-10","cited_by_count":280},{"id":"1605.06336","title":"Unsupervised Feature Extraction by Time-Contrastive Learning and Nonlinear ICA","created":"2016-05-20","cited_by_count":273},{"id":"1808.09031","title":"Targeted Syntactic Evaluation of Language Models","created":"2018-08-27","cited_by_count":242},{"id":"2307.12976","title":"Evaluating the Ripple Effects of Knowledge Editing in Language Models","created":"2023-07-24","cited_by_count":242},{"id":"2002.02886","title":"Weakly-Supervised Disentanglement Without Compromises","created":"2020-02-07","cited_by_count":239},{"id":"2501.17148","title":"AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders","created":"2025-01-28","cited_by_count":218},{"id":"2404.15255","title":"How to use and interpret activation patching","created":"2024-04-23","cited_by_count":214},{"id":"2405.14860","title":"Not All Language Model Features Are One-Dimensionally Linear","created":"2024-05-23","cited_by_count":212},{"id":"2303.02536","title":"Finding Alignments Between Interpretable Causal Variables and Distributed Neural Representations","created":"2023-03-05","cited_by_count":209},{"id":"2301.04709","title":"Causal Abstraction: A Theoretical Foundation for Mechanistic Interpretability","created":"2023-01-11","cited_by_count":205},{"id":"2404.16014","title":"Improving Dictionary Learning with Gated Sparse Autoencoders","created":"2024-04-24","cited_by_count":179},{"id":"2107.10098","title":"Disentanglement via Mechanism Sparsity Regularization: A New Principle for Nonlinear ICA","created":"2021-07-21","cited_by_count":160},{"id":"2307.15771","title":"The Hydra Effect: Emergent Self-repair in Language Model Computations","created":"2023-07-28","cited_by_count":156},{"id":"2005.00719","title":"Probing the Probing Paradigm: Does Probing Accuracy Entail Task Relevance?","created":"2020-05-02","cited_by_count":151},{"id":"2407.12404","title":"Analyzing the Generalization and Reliability of Steering Vectors","created":"2024-07-17","cited_by_count":143},{"id":"2209.11924","title":"Interventional Causal Representation Learning","created":"2022-09-24","cited_by_count":142},{"id":"2203.16437","title":"Weakly supervised causal representation learning","created":"2022-03-30","cited_by_count":140},{"id":"2202.03169","title":"CITRIS: Causal Identifiability from Temporal Intervened Sequences","created":"2022-02-07","cited_by_count":133},{"id":"1707.00819","title":"Causal Consistency of Structural Equation Models","created":"2017-07-04","cited_by_count":118},{"id":"2502.16681","title":"Are Sparse Autoencoders Useful? A Case Study in Sparse Probing","created":"2025-02-23","cited_by_count":115},{"id":"2410.13928","title":"Automatically Interpreting Millions of Features in Large Language Models","created":"2024-10-17","cited_by_count":113},{"id":"2112.00826","title":"Inducing Causal Structure for Interpretable Neural Networks","created":"2021-12-01","cited_by_count":110},{"id":"1812.03789","title":"Abstracting Causal Models","created":"2018-12-10","cited_by_count":107},{"id":"2007.00810","title":"On Linear Identifiability of Learned Representations","created":"2020-07-01","cited_by_count":105},{"id":"2002.11537","title":"ICE-BeeM: Identifiable Conditional Energy-Based Deep Models Based on Nonlinear ICA","created":"2020-02-26","cited_by_count":102},{"id":"2002.09815","title":"Neuron Shapley: Discovering the Responsible Neurons","created":"2020-02-23","cited_by_count":98},{"id":"2106.05200","title":"Independent mechanism analysis, a new concept?","created":"2021-06-09","cited_by_count":97},{"id":"2306.02235","title":"Learning Linear Causal Representations from Interventions under General Nonlinear Mixing","created":"2023-06-04","cited_by_count":92},{"id":"2503.10965","title":"Auditing language models for hidden objectives","created":"2025-03-14","cited_by_count":83},{"id":"2403.04121","title":"Can Large Language Models Reason and Plan?","created":"2024-03-07","cited_by_count":82},{"id":"2212.11870","title":"Impossibility Theorems for Feature Attribution","created":"2022-12-22","cited_by_count":78},{"id":"1503.01603","title":"External Validity: From Do-Calculus to Transportability Across Populations","created":"2015-03-05","cited_by_count":75},{"id":"2312.03813","title":"Improving Activation Steering in Language Models with Mean-Centring","created":"2023-12-06","cited_by_count":68},{"id":"2206.01101","title":"Weakly Supervised Representation Learning with Sparse Perturbations","created":"2022-06-02","cited_by_count":65},{"id":"2305.17553","title":"Detecting Edit Failures In Large Language Models: An Improved Specificity Benchmark","created":"2023-05-27","cited_by_count":62},{"id":"2307.02598","title":"Additive Decoders for Latent Variables Identification and Cartesian-Product Extrapolation","created":"2023-07-05","cited_by_count":58},{"id":"2504.13151","title":"MIB: A Mechanistic Interpretability Benchmark","created":"2025-04-17","cited_by_count":56},{"id":"2601.11516","title":"Building Production-Ready Probes For Gemini","created":"2026-01-16","cited_by_count":47},{"id":"2410.14670","title":"Decomposing The Dark Matter of Sparse Autoencoders","created":"2024-10-18","cited_by_count":45},{"id":"2203.15756","title":"Causal de Finetti: On the Identification of Invariant Causal Structure in Exchangeable Data","created":"2022-03-29","cited_by_count":42},{"id":"2310.16410","title":"Bridging the Human-AI Knowledge Gap: Concept Discovery and Transfer in AlphaZero","created":"2023-10-25","cited_by_count":42},{"id":"2502.09674","title":"The Hidden Dimensions of LLM Alignment: A Multi-Dimensional Analysis of Orthogonal Safety Directions","created":"2025-02-13","cited_by_count":41},{"id":"2409.04478","title":"Evaluating Open-Source Sparse Autoencoders on Disentangling Factual Knowledge in GPT-2 Small","created":"2024-09-05","cited_by_count":34},{"id":"2402.00849","title":"Score-based Causal Representation Learning: Linear and General Transformations","created":"2024-02-01","cited_by_count":27},{"id":"2407.04690","title":"Missed Causes and Ambiguous Effects: Counterfactuals Pose Challenges for Interpreting Neural Networks","created":"2024-07-05","cited_by_count":23},{"id":"2309.07742","title":"Interpretability is in the Mind of the Beholder: A Causal Framework for Human-interpretable Representation Learning","created":"2023-09-14","cited_by_count":22},{"id":"2406.14302","title":"Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning","created":"2024-06-20","cited_by_count":16},{"id":"2505.11756","title":"Feature Hedging: Correlated Features Break Narrow Sparse Autoencoders","created":"2025-05-16","cited_by_count":16},{"id":"2211.09169","title":"Engineering Monosemanticity in Toy Models","created":"2022-11-16","cited_by_count":13},{"id":"2406.13371","title":"Identifiable Causal Representation Learning: Unsupervised, Multi-View, and Multi-Environment","created":"2024-06-19","cited_by_count":12},{"id":"2512.15134","title":"From Isolation to Entanglement: When Do Interpretability Methods Identify and Disentangle Known Concepts?","created":"2025-12-17","cited_by_count":7},{"id":"2509.23323","title":"LLM Interpretability with Identifiable Temporal-Instantaneous Representation","created":"2025-09-27","cited_by_count":6},{"id":"2510.08506","title":"Neologism Learning for Controllability and Self-Verbalization","created":"2025-10-09","cited_by_count":6},{"id":"2502.05085","title":"Causality can systematically address the monsters under the bench(marks)","created":"2025-02-07","cited_by_count":4},{"id":"2601.12913","title":"Actionable Interpretability Must Be Defined in Terms of Symmetries","created":"2026-01-19","cited_by_count":4},{"id":"2602.04718","title":"Superposition Without Interference? Towards Isolated Interventions via Almost Orthogonal Features in Language Models","created":"2026-02-04","cited_by_count":2}],"next_cursor":null}