{"total":187,"rows":[{"id":"2303.12712","title":"Sparks of Artificial General Intelligence: Early experiments with GPT-4","created":"2023-03-22","cited_by_count":2592,"area":null},{"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,"area":"mechinterp"},{"id":"2301.05217","title":"Progress measures for grokking via mechanistic interpretability","created":"2023-01-12","cited_by_count":839,"area":"mechinterp"},{"id":"2307.05782","title":"Large Language Models","created":"2023-07-11","cited_by_count":780,"area":null},{"id":"2304.15004","title":"Are Emergent Abilities of Large Language Models a Mirage?","created":"2023-04-28","cited_by_count":523,"area":null},{"id":"2404.14082","title":"Mechanistic Interpretability for AI Safety -- A Review","created":"2024-04-22","cited_by_count":422,"area":"mechinterp"},{"id":"2309.16042","title":"Towards Best Practices of Activation Patching in Language Models: Metrics and Methods","created":"2023-09-27","cited_by_count":359,"area":"mechinterp"},{"id":"2210.10749","title":"Transformers Learn Shortcuts to Automata","created":"2022-10-19","cited_by_count":268,"area":null},{"id":"2306.17844","title":"The Clock and the Pizza: Two Stories in Mechanistic Explanation of Neural Networks","created":"2023-06-30","cited_by_count":156,"area":null},{"id":"2210.01117","title":"Omnigrok: Grokking Beyond Algorithmic Data","created":"2022-10-03","cited_by_count":146,"area":null},{"id":"2302.03025","title":"A Toy Model of Universality: Reverse Engineering How Networks Learn Group Operations","created":"2023-02-06","cited_by_count":144,"area":"mechinterp"},{"id":"2302.11055","title":"SGD learning on neural networks: leap complexity and saddle-to-saddle dynamics","created":"2023-02-21","cited_by_count":143,"area":null},{"id":"2303.13506","title":"The Quantization Model of Neural Scaling","created":"2023-03-23","cited_by_count":138,"area":null},{"id":"2309.07311","title":"Sudden Drops in the Loss: Syntax Acquisition, Phase Transitions, and Simplicity Bias in MLMs","created":"2023-09-13","cited_by_count":124,"area":null},{"id":"2402.11004","title":"The Evolution of Statistical Induction Heads: In-Context Learning Markov Chains","created":"2024-02-16","cited_by_count":118,"area":null},{"id":"2309.02390","title":"Explaining grokking through circuit efficiency","created":"2023-09-05","cited_by_count":111,"area":null},{"id":"2310.06110","title":"Grokking as the Transition from Lazy to Rich Training Dynamics","created":"2023-10-09","cited_by_count":108,"area":null},{"id":"2402.04248","title":"Can Mamba Learn How to Learn? A Comparative Study on In-Context Learning Tasks","created":"2024-02-06","cited_by_count":105,"area":"ssm"},{"id":"2305.16380","title":"Scan and Snap: Understanding Training Dynamics and Token Composition in 1-layer Transformer","created":"2023-05-25","cited_by_count":102,"area":null},{"id":"2411.04330","title":"Scaling Laws for Precision","created":"2024-11-07","cited_by_count":95,"area":null},{"id":"2310.09336","title":"Compositional Abilities Emerge Multiplicatively: Exploring Diffusion Models on a Synthetic Task","created":"2023-10-13","cited_by_count":93,"area":"diffusion"},{"id":"2301.02679","title":"Grokking modular arithmetic","created":"2023-01-06","cited_by_count":83,"area":null},{"id":"2303.11873","title":"A Tale of Two Circuits: Grokking as Competition of Sparse and Dense Subnetworks","created":"2023-03-21","cited_by_count":83,"area":null},{"id":"2311.18817","title":"Dichotomy of Early and Late Phase Implicit Biases Can Provably Induce Grokking","created":"2023-11-30","cited_by_count":80,"area":null},{"id":"2306.00946","title":"Exposing Attention Glitches with Flip-Flop Language Modeling","created":"2023-06-01","cited_by_count":75,"area":null},{"id":"2307.02129","title":"How Deep Neural Networks Learn Compositional Data: The Random Hierarchy Model","created":"2023-07-05","cited_by_count":71,"area":null},{"id":"2211.12316","title":"Simplicity Bias in Transformers and their Ability to Learn Sparse Boolean Functions","created":"2022-11-22","cited_by_count":70,"area":null},{"id":"2210.05337","title":"SGD with Large Step Sizes Learns Sparse Features","created":"2022-10-11","cited_by_count":68,"area":null},{"id":"2209.14863","title":"Neural Networks Efficiently Learn Low-Dimensional Representations with SGD","created":"2022-09-29","cited_by_count":63,"area":null},{"id":"2303.00055","title":"Learning time-scales in two-layers neural networks","created":"2023-02-28","cited_by_count":57,"area":null},{"id":"2405.11751","title":"Asymptotic theory of in-context learning by linear attention","created":"2024-05-20","cited_by_count":55,"area":null},{"id":"2310.00535","title":"JoMA: Demystifying Multilayer Transformers via JOint Dynamics of MLP and Attention","created":"2023-10-01","cited_by_count":52,"area":null},{"id":"2402.15555","title":"Deep Networks Always Grok and Here is Why","created":"2024-02-23","cited_by_count":51,"area":null},{"id":"2501.04697","title":"Grokking at the Edge of Numerical Stability","created":"2025-01-08","cited_by_count":46,"area":null},{"id":"2303.16504","title":"An Over-parameterized Exponential Regression","created":"2023-03-29","cited_by_count":45,"area":null},{"id":"2405.19592","title":"Why Larger Language Models Do In-context Learning Differently?","created":"2024-05-30","cited_by_count":45,"area":null},{"id":"2409.01659","title":"Interpreting and Improving Large Language Models in Arithmetic Calculation","created":"2024-09-03","cited_by_count":45,"area":null},{"id":"2405.20233","title":"Grokfast: Accelerated Grokking by Amplifying Slow Gradients","created":"2024-05-30","cited_by_count":41,"area":null},{"id":"2402.02364","title":"Loss Landscape Degeneracy and Stagewise Development in Transformers","created":"2024-02-04","cited_by_count":40,"area":null},{"id":"2309.15111","title":"SGD Finds then Tunes Features in Two-Layer Neural Networks with near-Optimal Sample Complexity: A Case Study in the XOR problem","created":"2023-09-26","cited_by_count":39,"area":null},{"id":"2407.12332","title":"Why Do You Grok? A Theoretical Analysis of Grokking Modular Addition","created":"2024-07-17","cited_by_count":36,"area":null},{"id":"2406.19370","title":"Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Space","created":"2024-06-27","cited_by_count":35,"area":null},{"id":"2309.08632","title":"Pretraining on the Test Set Is All You Need","created":"2023-09-13","cited_by_count":33,"area":null},{"id":"2310.02541","title":"Benign Overfitting and Grokking in ReLU Networks for XOR Cluster Data","created":"2023-10-04","cited_by_count":33,"area":null},{"id":"2311.07568","title":"Feature emergence via margin maximization: case studies in algebraic tasks","created":"2023-11-13","cited_by_count":33,"area":null},{"id":"2309.03843","title":"Gradient-Based Feature Learning under Structured Data","created":"2023-09-07","cited_by_count":32,"area":null},{"id":"2406.02128","title":"Iteration Head: A Mechanistic Study of Chain-of-Thought","created":"2024-06-04","cited_by_count":31,"area":"lr"},{"id":"2408.11804","title":"Approaching Deep Learning through the Spectral Dynamics of Weights","created":"2024-08-21","cited_by_count":31,"area":null},{"id":"2411.02544","title":"Pretrained transformer efficiently learns low-dimensional target functions in-context","created":"2024-11-04","cited_by_count":31,"area":null},{"id":"2310.06301","title":"Dynamical versus Bayesian Phase Transitions in a Toy Model of Superposition","created":"2023-10-10","cited_by_count":29,"area":"mechinterp"},{"id":"2410.05459","title":"From Sparse Dependence to Sparse Attention: Unveiling How Chain-of-Thought Enhances Transformer Sample Efficiency","created":"2024-10-07","cited_by_count":29,"area":"lr"},{"id":"2407.20199","title":"Emergence in non-neural models: grokking modular arithmetic via average gradient outer product","created":"2024-07-29","cited_by_count":28,"area":null},{"id":"2405.04669","title":"Towards a Theoretical Understanding of the 'Reversal Curse' via Training Dynamics","created":"2024-05-07","cited_by_count":27,"area":"lr"},{"id":"2310.13061","title":"To grok or not to grok: Disentangling generalization and memorization on corrupted algorithmic datasets","created":"2023-10-19","cited_by_count":24,"area":null},{"id":"2312.06581","title":"Grokking Group Multiplication with Cosets","created":"2023-12-11","cited_by_count":24,"area":null},{"id":"2306.16921","title":"Provable Advantage of Curriculum Learning on Parity Targets with Mixed Inputs","created":"2023-06-29","cited_by_count":22,"area":null},{"id":"2301.01597","title":"Problem-Dependent Power of Quantum Neural Networks on Multi-Class Classification","created":"2022-12-29","cited_by_count":20,"area":null},{"id":"2301.13833","title":"A Mathematical Model for Curriculum Learning for Parities","created":"2023-01-31","cited_by_count":19,"area":null},{"id":"2305.06986","title":"Provable Guarantees for Nonlinear Feature Learning in Three-Layer Neural Networks","created":"2023-05-11","cited_by_count":19,"area":null},{"id":"2406.03495","title":"Grokking Modular Polynomials","created":"2024-06-05","cited_by_count":18,"area":null},{"id":"2307.15804","title":"On Single Index Models beyond Gaussian Data","created":"2023-07-28","cited_by_count":17,"area":null},{"id":"2408.08944","title":"Information-Theoretic Progress Measures reveal Grokking is an Emergent Phase Transition","created":"2024-08-16","cited_by_count":17,"area":null},{"id":"2307.06887","title":"Provable Multi-Task Representation Learning by Two-Layer ReLU Neural Networks","created":"2023-07-13","cited_by_count":16,"area":null},{"id":"2308.09543","title":"Latent State Models of Training Dynamics","created":"2023-08-18","cited_by_count":16,"area":null},{"id":"2506.18110","title":"RL for Reasoning by Adaptively Revealing Rationales","created":"2025-06-22","cited_by_count":16,"area":null},{"id":"2403.09053","title":"Towards a theory of model distillation","created":"2024-03-14","cited_by_count":15,"area":null},{"id":"2502.20548","title":"$Q\\sharp$: Provably Optimal Distributional RL for LLM Post-Training","created":"2025-02-27","cited_by_count":15,"area":"lr"},{"id":"2509.21519","title":"Provable Scaling Laws of Feature Emergence from Learning Dynamics of Grokking","created":"2025-09-25","cited_by_count":15,"area":null},{"id":"2208.02789","title":"Feature selection with gradient descent on two-layer networks in low-rotation regimes","created":"2022-08-04","cited_by_count":14,"area":null},{"id":"2310.05918","title":"Grokking as Compression: A Nonlinear Complexity Perspective","created":"2023-10-09","cited_by_count":14,"area":null},{"id":"2310.11428","title":"Butterfly Effects of SGD Noise: Error Amplification in Behavior Cloning and Autoregression","created":"2023-10-17","cited_by_count":14,"area":null},{"id":"2310.12408","title":"Provable Guarantees for Neural Networks via Gradient Feature Learning","created":"2023-10-19","cited_by_count":14,"area":null},{"id":"2310.12956","title":"Eureka-Moments in Transformers: Multi-Step Tasks Reveal Softmax Induced Optimization Problems","created":"2023-10-19","cited_by_count":14,"area":null},{"id":"2402.16726","title":"Towards Empirical Interpretation of Internal Circuits and Properties in Grokked Transformers on Modular Polynomials","created":"2024-02-26","cited_by_count":14,"area":null},{"id":"2403.06925","title":"Transformers Learn Low Sensitivity Functions: Investigations and Implications","created":"2024-03-11","cited_by_count":14,"area":null},{"id":"2502.01481","title":"Intrinsic Entropy of Context Length Scaling in LLMs","created":"2025-02-03","cited_by_count":14,"area":null},{"id":"2506.06489","title":"Alternating Gradient Flows: A Theory of Feature Learning in Two-layer Neural Networks","created":"2025-06-06","cited_by_count":14,"area":null},{"id":"2309.03800","title":"Pareto Frontiers in Neural Feature Learning: Data, Compute, Width, and Luck","created":"2023-09-07","cited_by_count":13,"area":null},{"id":"2310.17247","title":"Grokking Beyond Neural Networks: An Empirical Exploration with Model Complexity","created":"2023-10-26","cited_by_count":13,"area":null},{"id":"2310.17639","title":"In-Context Learning Dynamics with Random Binary Sequences","created":"2023-10-26","cited_by_count":13,"area":null},{"id":"2311.04163","title":"Outliers with Opposing Signals Have an Outsized Effect on Neural Network Optimization","created":"2023-11-07","cited_by_count":13,"area":null},{"id":"2504.13292","title":"Let Me Grok for You: Accelerating Grokking via Embedding Transfer from a Weaker Model","created":"2025-04-17","cited_by_count":13,"area":null},{"id":"2505.20172","title":"A Theoretical Framework for Grokking: Interpolation followed by Riemannian Norm Minimisation","created":"2025-05-26","cited_by_count":13,"area":null},{"id":"2211.11052","title":"Convexifying Transformers: Improving optimization and understanding of transformer networks","created":"2022-11-20","cited_by_count":12,"area":null},{"id":"2311.06597","title":"Understanding Grokking Through A Robustness Viewpoint","created":"2023-11-11","cited_by_count":12,"area":null},{"id":"2403.03375","title":"Complexity Matters: Dynamics of Feature Learning in the Presence of Spurious Correlations","created":"2024-03-05","cited_by_count":12,"area":null},{"id":"2506.05295","title":"Sample Complexity and Representation Ability of Test-time Scaling Paradigms","created":"2025-06-05","cited_by_count":12,"area":"lr"},{"id":"2310.07325","title":"An Adversarial Example for Direct Logit Attribution: Memory Management in GELU-4L","created":"2023-10-11","cited_by_count":11,"area":null},{"id":"2410.22244","title":"Abrupt Learning in Transformers: A Case Study on Matrix Completion","created":"2024-10-29","cited_by_count":11,"area":null},{"id":"2402.05271","title":"Feature learning as alignment: a structural property of gradient descent in non-linear neural networks","created":"2024-02-07","cited_by_count":10,"area":null},{"id":"2403.03942","title":"The Heuristic Core: Understanding Subnetwork Generalization in Pretrained Language Models","created":"2024-03-06","cited_by_count":10,"area":null},{"id":"2412.04619","title":"Sometimes I am a Tree: Data Drives Unstable Hierarchical Generalization","created":"2024-12-05","cited_by_count":10,"area":null},{"id":"2602.03655","title":"Sequential Group Composition: A Window into the Mechanics of Deep Learning","created":"2026-02-03","cited_by_count":10,"area":null},{"id":"2411.08798","title":"Learning Gaussian Multi-Index Models with Gradient Flow: Time Complexity and Directional Convergence","created":"2024-11-13","cited_by_count":9,"area":null},{"id":"2608.18222","title":"Think Shallow, Solve Deep: Controlling Recurrent Dynamics for Reliable Test-Time Depth","created":"2026-08-18","cited_by_count":9,"area":null},{"id":"2405.12755","title":"Progress Measures for Grokking on Real-world Tasks","created":"2024-05-21","cited_by_count":8,"area":null},{"id":"2502.06443","title":"Low-dimensional Functions are Efficiently Learnable under Randomly Biased Distributions","created":"2025-02-10","cited_by_count":8,"area":null},{"id":"2601.19791","title":"To Grok Grokking: Provable Grokking in Ridge Regression","created":"2026-01-27","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":"2410.10024","title":"Sharper Guarantees for Learning Neural Network Classifiers with Gradient Methods","created":"2024-10-13","cited_by_count":7,"area":null}],"next_cursor":"100"}