{"total":63,"rows":[{"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":"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":"2410.09087","title":"Mechanistic?","created":"2024-10-07","cited_by_count":25,"area":"mechinterp"},{"id":"2502.11196","title":"How Do LLMs Acquire New Knowledge? A Knowledge Circuits Perspective on Continual Pre-Training","created":"2025-02-16","cited_by_count":23,"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":"2410.01444","title":"Geometric Signatures of Compositionality Across a Language Model's Lifetime","created":"2024-10-02","cited_by_count":16,"area":null},{"id":"2507.10616","title":"Scalpel vs. Hammer: GRPO Amplifies Existing Capabilities, SFT Replaces Them","created":"2025-07-13","cited_by_count":15,"area":"lr"},{"id":"2601.21996","title":"Mechanistic Data Attribution: Tracing the Training Origins of Interpretable LLM Units","created":"2026-01-29","cited_by_count":10,"area":"mechinterp"},{"id":"2505.15105","title":"Mechanistic evaluation of Transformers and state space models","created":"2025-05-21","cited_by_count":9,"area":"ssm"},{"id":"2507.10624","title":"Comprehension Without Competence: Architectural Limits of LLMs in Symbolic Computation and Reasoning","created":"2025-07-14","cited_by_count":9,"area":"lr"},{"id":"2510.12071","title":"Influence Dynamics and Stagewise Data Attribution","created":"2025-10-14","cited_by_count":7,"area":null},{"id":"2505.24731","title":"Circuit Stability Characterizes Language Model Generalization","created":"2025-05-30","cited_by_count":6,"area":"mechinterp"},{"id":"2509.05291","title":"Crosscoding Through Time: Tracking Emergence & Consolidation Of Linguistic Representations Throughout LLM Pretraining","created":"2025-09-05","cited_by_count":6,"area":null},{"id":"2504.15471","title":"Bigram Subnetworks: Mapping to Next Tokens in Transformer Language Models","created":"2025-04-21","cited_by_count":5,"area":null},{"id":"2505.20225","title":"FLAME-MoE: A Transparent End-to-End Research Platform for Mixture-of-Experts Language Models","created":"2025-05-26","cited_by_count":5,"area":"moe"},{"id":"2509.22831","title":"Toward a Theory of Generalizability in LLM Mechanistic Interpretability Research","created":"2025-09-26","cited_by_count":5,"area":"mechinterp"},{"id":"2410.08255","title":"Investigating Representation Universality: Case Study on Genealogical Representations","created":"2024-10-10","cited_by_count":4,"area":null},{"id":"2505.20076","title":"ExPLAIND: Unifying Model, Data, and Training Attribution to Study Model Behavior","created":"2025-05-26","cited_by_count":3,"area":null},{"id":"2506.13727","title":"Attribution-Guided Pruning for Insight and Control: Circuit Discovery and Targeted Correction in Small-scale LLMs","created":"2025-06-16","cited_by_count":3,"area":"mechinterp"},{"id":"2509.26643","title":"Convergence and Divergence of Language Models under Different Random Seeds","created":"2025-09-30","cited_by_count":3,"area":null},{"id":"2511.21974","title":"Start Making Sense(s): A Developmental Probe of Attention Specialization Using Lexical Ambiguity","created":"2025-11-26","cited_by_count":3,"area":null},{"id":"2601.18302","title":"Suppressing Final Layer Hidden State Jumps in Transformer Pretraining","created":"2026-01-26","cited_by_count":3,"area":null},{"id":"2601.22510","title":"Shattered Compositionality: Counterintuitive Learning Dynamics of Transformers for Arithmetic","created":"2026-01-30","cited_by_count":3,"area":null},{"id":"2602.03484","title":"Preferences for Idiomatic Language are Acquired Slowly -- and Forgotten Quickly: A Case Study on Swedish","created":"2026-02-03","cited_by_count":3,"area":null},{"id":"2602.19058","title":"Do LLMs and VLMs Share Neurons for Inference? Evidence and Mechanisms of Cross-Modal Transfer","created":"2026-02-22","cited_by_count":3,"area":"mllm"},{"id":"2604.24302","title":"Differentiable Faithfulness Alignment for Cross-Model Circuit Transfer","created":"2026-04-27","cited_by_count":3,"area":"mechinterp"},{"id":"2606.06533","title":"Position: Don't Just \"Fix it in Post\": A Science of AI Must Study Training Dynamics","created":"2026-06-03","cited_by_count":3,"area":null},{"id":"2503.05760","title":"The Lazy Student's Dream: ChatGPT Passing an Engineering Course on Its Own","created":"2025-02-23","cited_by_count":2,"area":null},{"id":"2509.15560","title":"How important is language for human-like intelligence?","created":"2025-09-19","cited_by_count":2,"area":null},{"id":"2510.00034","title":"Review of Hallucination Understanding in Large Language and Vision Models","created":"2025-09-26","cited_by_count":2,"area":null},{"id":"2511.20910","title":"Emergence and Localisation of Semantic Role Circuits in LLMs","created":"2025-11-25","cited_by_count":2,"area":"mechinterp"},{"id":"2603.03993","title":"Specialization of softmax attention heads: insights from the high-dimensional single-location model","created":"2026-03-04","cited_by_count":2,"area":null},{"id":"2603.20101","title":"Pitfalls in Evaluating Interpretability Agents","created":"2026-03-20","cited_by_count":2,"area":"agents"},{"id":"2605.03354","title":"What Happens Inside Agent Memory? Circuit Analysis from Emergence to Diagnosis","created":"2026-05-05","cited_by_count":2,"area":"agents"},{"id":"2606.00033","title":"Make Mechanistic Interpretability Auditable: A Call to Develop Guidelines via Continuous Collaborative Reviewing","created":"2026-04-24","cited_by_count":2,"area":"mechinterp"},{"id":"2607.18921","title":"Circuit Claims Depend on What Is Extracted and How It Is Compared","created":"2026-07-21","cited_by_count":2,"area":null},{"id":"2411.07071","title":"Universal Response and Emergence of Induction in LLMs","created":"2024-11-11","cited_by_count":1,"area":null},{"id":"2502.20475","title":"Promote, Suppress, Iterate: How Language Models Answer One-to-Many Factual Queries","created":"2025-02-27","cited_by_count":1,"area":null},{"id":"2508.10553","title":"eDIF: A European Deep Inference Fabric for Remote Interpretability of LLM","created":"2025-08-14","cited_by_count":1,"area":"mechinterp"},{"id":"2601.03570","title":"How Do Large Language Models Learn Concepts During Continual Pre-Training?","created":"2026-01-07","cited_by_count":1,"area":null},{"id":"2602.13524","title":"Singular Vectors of Attention Heads Align with Features","created":"2026-02-13","cited_by_count":1,"area":"mechinterp"},{"id":"2604.15294","title":"How Do LLMs and VLMs Understand Viewpoint Rotation Without Vision? An Interpretability Study","created":"2026-04-16","cited_by_count":1,"area":null},{"id":"2605.16023","title":"Judge Circuits Explain Format-Induced Inconsistency in LLM-as-a-Judge","created":"2026-05-15","cited_by_count":1,"area":null},{"id":"2607.06648","title":"Final Checkpoints Are Not Enough: Analyzing Latent Reasoning Faithfulness Along Training Trajectories","created":"2026-07-07","cited_by_count":1,"area":null},{"id":"2609.03887","title":"Beyond Shallow Alignment: How Post-Training Methods Determine Refusal Circuits And Steering Robustness","created":"2026-09-03","cited_by_count":1,"area":"alignment"},{"id":"2503.00224","title":"\\`A la recherche du sens perdu: your favourite LLM might have more to say than you can understand","created":"2025-02-28","cited_by_count":0,"area":null},{"id":"2506.03434","title":"Time Course MechInterp: Analyzing the Evolution of Components and Knowledge in Large Language Models","created":"2025-06-03","cited_by_count":0,"area":null},{"id":"2601.14758","title":"Mechanism Shift During Post-training from Autoregressive to Masked Diffusion Language Models","created":"2026-01-21","cited_by_count":0,"area":"diffusion"},{"id":"2602.12811","title":"Left-right asymmetry in predicting brain activity from LLMs' representations emerges with their formal linguistic competence","created":"2026-02-13","cited_by_count":0,"area":null},{"id":"2603.14087","title":"Understanding the Emergence of Seemingly Useless Features in Next-Token Predictors","created":"2026-03-14","cited_by_count":0,"area":null},{"id":"2606.28524","title":"Developmental Trajectories of Situation Modeling and Mentalizing in Transformer Language Models","created":"2026-06-26","cited_by_count":0,"area":null},{"id":"2607.06621","title":"Fingerprint, Not Blueprint: How Positional Schemes Set the Default Spectral Algebra of Attention","created":"2026-07-07","cited_by_count":0,"area":null},{"id":"2608.05158","title":"Safe Evolution with Circuit Anchors","created":"2026-05-24","cited_by_count":0,"area":null},{"id":"2608.15483","title":"Measuring Structured Predictability in Neural Training Dynamics: A Cross-Regime Study","created":"2026-08-16","cited_by_count":0,"area":null},{"id":"2608.16085","title":"Behaviour Is an Incomplete Measure of Reasoning Development: Cross-surface pre-arrival accessibility and the limits of developmental inference in a recurrent-depth reasoner","created":"2026-08-17","cited_by_count":0,"area":null},{"id":"2608.27254","title":"Circuit Condensation: Post-Training that Concentrates a Behavior's Causal Circuit","created":"2026-08-27","cited_by_count":0,"area":"mechinterp"},{"id":"2609.24352","title":"Few-Shot Demonstrations Elicit the Use of In-Context World Representations in LLMs","created":"2026-09-21","cited_by_count":0,"area":null},{"id":"2609.35663","title":"Late Attention Layers Alone Can Copy Entity Tokens, but Not Without Attending to Their Context","created":"2026-09-28","cited_by_count":0,"area":null},{"id":"2609.35890","title":"Representational Simplicity and Circuit Size Dissociate in a Threshold-Dependent Way: A Controlled Test via Adversarial Training","created":"2026-09-27","cited_by_count":0,"area":"mechinterp"},{"id":"2610.01165","title":"Persistent Depth Ordering amid Shifting Block-Bypass Responses in Language Model Pretraining","created":"2026-10-01","cited_by_count":0,"area":null},{"id":"2610.04119","title":"Copying Before Suppression: What Drives a Below-Chance Dip During Language Model Training?","created":"2026-10-02","cited_by_count":0,"area":"mechinterp"},{"id":"2610.19113","title":"From Copying to Composition: How Language Models Develop In-Context Learning","created":"2026-10-13","cited_by_count":0,"area":null},{"id":"2610.19907","title":"Sparse Transformer: a Controlled Testbed for Mechanistic Interpretability","created":"2026-10-14","cited_by_count":0,"area":"mechinterp"}],"next_cursor":null}