{"total":190,"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":"2307.05782","title":"Large Language Models","created":"2023-07-11","cited_by_count":780,"area":null},{"id":"2305.18654","title":"Faith and Fate: Limits of Transformers on Compositionality","created":"2023-05-29","cited_by_count":503,"area":null},{"id":"2306.09927","title":"Trained Transformers Learn Linear Models In-Context","created":"2023-06-16","cited_by_count":337,"area":null},{"id":"2306.04637","title":"Transformers as Statisticians: Provable In-Context Learning with In-Context Algorithm Selection","created":"2023-06-07","cited_by_count":298,"area":null},{"id":"2301.13196","title":"Looped Transformers as Programmable Computers","created":"2023-01-30","cited_by_count":237,"area":null},{"id":"2404.08819","title":"The Illusion of State in State-Space Models","created":"2024-04-12","cited_by_count":187,"area":null},{"id":"2307.03576","title":"One Step of Gradient Descent is Provably the Optimal In-Context Learner with One Layer of Linear Self-Attention","created":"2023-07-07","cited_by_count":170,"area":null},{"id":"2306.00802","title":"Birth of a Transformer: A Memory Viewpoint","created":"2023-06-01","cited_by_count":161,"area":null},{"id":"2403.06963","title":"The pitfalls of next-token prediction","created":"2024-03-11","cited_by_count":160,"area":null},{"id":"2311.12424","title":"Looped Transformers are Better at Learning Learning Algorithms","created":"2023-11-21","cited_by_count":149,"area":null},{"id":"2306.02896","title":"Representational Strengths and Limitations of Transformers","created":"2023-06-05","cited_by_count":133,"area":null},{"id":"2311.00208","title":"What Formal Languages Can Transformers Express? A Survey","created":"2023-11-01","cited_by_count":124,"area":null},{"id":"2311.12786","title":"Mechanistically analyzing the effects of fine-tuning on procedurally defined tasks","created":"2023-11-21","cited_by_count":115,"area":null},{"id":"2309.05858","title":"Uncovering mesa-optimization algorithms in Transformers","created":"2023-09-11","cited_by_count":100,"area":null},{"id":"2303.04245","title":"How Do Transformers Learn Topic Structure: Towards a Mechanistic Understanding","created":"2023-03-07","cited_by_count":86,"area":null},{"id":"2402.09371","title":"Transformers Can Achieve Length Generalization But Not Robustly","created":"2024-02-14","cited_by_count":86,"area":null},{"id":"2402.09268","title":"Transformers, parallel computation, and logarithmic depth","created":"2024-02-14","cited_by_count":79,"area":null},{"id":"2405.15943","title":"Transformers represent belief state geometry in their residual stream","created":"2024-05-24","cited_by_count":77,"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":"2502.10297","title":"DeltaProduct: Improving State-Tracking in Linear RNNs via Householder Products","created":"2025-02-14","cited_by_count":75,"area":null},{"id":"2402.18510","title":"RNNs are not Transformers (Yet): The Key Bottleneck on In-context Retrieval","created":"2024-02-28","cited_by_count":64,"area":null},{"id":"2304.05366","title":"The No Free Lunch Theorem, Kolmogorov Complexity, and the Role of Inductive Biases in Machine Learning","created":"2023-04-11","cited_by_count":63,"area":null},{"id":"2310.03016","title":"Understanding In-Context Learning in Transformers and LLMs by Learning to Learn Discrete Functions","created":"2023-10-04","cited_by_count":63,"area":null},{"id":"2503.03961","title":"A Little Depth Goes a Long Way: The Expressive Power of Log-Depth Transformers","created":"2025-03-05","cited_by_count":53,"area":null},{"id":"2501.00070","title":"ICLR: In-Context Learning of Representations","created":"2024-12-29","cited_by_count":50,"area":null},{"id":"2303.08117","title":"Do Transformers Parse while Predicting the Masked Word?","created":"2023-03-14","cited_by_count":48,"area":null},{"id":"2405.17394","title":"The Expressive Capacity of State Space Models: A Formal Language Perspective","created":"2024-05-27","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":"2412.02975","title":"Theoretical limitations of multi-layer Transformer","created":"2024-12-04","cited_by_count":45,"area":null},{"id":"2307.11353","title":"What can a Single Attention Layer Learn? A Study Through the Random Features Lens","created":"2023-07-21","cited_by_count":40,"area":null},{"id":"2401.15530","title":"An Information-Theoretic Analysis of In-Context Learning","created":"2024-01-28","cited_by_count":40,"area":null},{"id":"2503.02854","title":"(How) Do Language Models Track State?","created":"2025-03-04","cited_by_count":34,"area":null},{"id":"2505.16381","title":"PaTH Attention: Position Encoding via Accumulating Householder Transformations","created":"2025-05-22","cited_by_count":34,"area":null},{"id":"2401.14953","title":"Learning Universal Predictors","created":"2024-01-26","cited_by_count":31,"area":null},{"id":"2312.01429","title":"Transformers are uninterpretable with myopic methods: a case study with bounded Dyck grammars","created":"2023-12-03","cited_by_count":30,"area":null},{"id":"2402.01107","title":"Simulation of Graph Algorithms with Looped Transformers","created":"2024-02-02","cited_by_count":30,"area":null},{"id":"2405.16411","title":"Training Tensor Attention Efficiently: From Cubic to Almost Linear Time","created":"2024-05-26","cited_by_count":30,"area":null},{"id":"2406.09347","title":"Separations in the Representational Capabilities of Transformers and Recurrent Architectures","created":"2024-06-13","cited_by_count":29,"area":null},{"id":"2504.15266","title":"Roll the dice & look before you leap: Going beyond the creative limits of next-token prediction","created":"2025-04-21","cited_by_count":29,"area":null},{"id":"2305.15287","title":"The Crucial Role of Normalization in Sharpness-Aware Minimization","created":"2023-05-24","cited_by_count":26,"area":null},{"id":"2308.03212","title":"Average-Hard Attention Transformers are Constant-Depth Uniform Threshold Circuits","created":"2023-08-06","cited_by_count":26,"area":null},{"id":"2408.12578","title":"A Percolation Model of Emergence: Analyzing Transformers Trained on a Formal Language","created":"2024-08-22","cited_by_count":26,"area":null},{"id":"2411.07602","title":"Circuit Complexity Bounds for RoPE-based Transformer Architecture","created":"2024-11-12","cited_by_count":26,"area":null},{"id":"2305.16704","title":"A Closer Look at In-Context Learning under Distribution Shifts","created":"2023-05-26","cited_by_count":25,"area":null},{"id":"2312.03656","title":"Interpretability Illusions in the Generalization of Simplified Models","created":"2023-12-06","cited_by_count":25,"area":null},{"id":"2402.00522","title":"Understanding the Expressive Power and Mechanisms of Transformer for Sequence Modeling","created":"2024-02-01","cited_by_count":25,"area":null},{"id":"2405.20131","title":"Language Models Need Inductive Biases to Count Inductively","created":"2024-05-30","cited_by_count":25,"area":null},{"id":"2407.03310","title":"Universal Length Generalization with Turing Programs","created":"2024-07-03","cited_by_count":25,"area":null},{"id":"2401.16421","title":"Two Stones Hit One Bird: Bilevel Positional Encoding for Better Length Extrapolation","created":"2024-01-29","cited_by_count":24,"area":null},{"id":"2407.10005","title":"Fine-grained Analysis of In-context Linear Estimation: Data, Architecture, and Beyond","created":"2024-07-13","cited_by_count":24,"area":null},{"id":"2408.07144","title":"Language Models as Models of Language","created":"2024-08-13","cited_by_count":24,"area":null},{"id":"2311.04128","title":"Generative learning for nonlinear dynamics","created":"2023-11-07","cited_by_count":22,"area":null},{"id":"2311.12997","title":"Compositional Capabilities of Autoregressive Transformers: A Study on Synthetic, Interpretable Tasks","created":"2023-11-21","cited_by_count":22,"area":null},{"id":"2402.11639","title":"In-Context Learning with Transformers: Softmax Attention Adapts to Function Lipschitzness","created":"2024-02-18","cited_by_count":22,"area":null},{"id":"2406.06893","title":"Transformers Provably Learn Sparse Token Selection While Fully-Connected Nets Cannot","created":"2024-06-11","cited_by_count":22,"area":null},{"id":"2402.13934","title":"Do Efficient Transformers Really Save Computation?","created":"2024-02-21","cited_by_count":20,"area":null},{"id":"2404.01601","title":"What Can Transformer Learn with Varying Depth? Case Studies on Sequence Learning Tasks","created":"2024-04-02","cited_by_count":19,"area":null},{"id":"2405.16674","title":"Limits of Deep Learning: Sequence Modeling through the Lens of Complexity Theory","created":"2024-05-26","cited_by_count":19,"area":null},{"id":"2410.09375","title":"Looped ReLU MLPs May Be All You Need as Practical Programmable Computers","created":"2024-10-12","cited_by_count":19,"area":null},{"id":"2310.08540","title":"Do pretrained Transformers Learn In-Context by Gradient Descent?","created":"2023-10-12","cited_by_count":18,"area":null},{"id":"2505.17761","title":"Structured Linear CDEs: Maximally Expressive and Parallel-in-Time Sequence Models","created":"2025-05-23","cited_by_count":18,"area":null},{"id":"2305.14699","title":"Can Transformers Learn to Solve Problems Recursively?","created":"2023-05-24","cited_by_count":17,"area":null},{"id":"2307.01189","title":"Trainable Transformer in Transformer","created":"2023-07-03","cited_by_count":17,"area":null},{"id":"2405.05409","title":"Initialization is Critical to Whether Transformers Fit Composite Functions by Reasoning or Memorizing","created":"2024-05-08","cited_by_count":17,"area":null},{"id":"2509.22284","title":"Structured Sparse Transition Matrices to Enable State Tracking in State-Space Models","created":"2025-09-26","cited_by_count":16,"area":null},{"id":"2410.05464","title":"Progressive distillation induces an implicit curriculum","created":"2024-10-07","cited_by_count":15,"area":null},{"id":"2412.07942","title":"Neural Scaling Laws Rooted in the Data Distribution","created":"2024-12-10","cited_by_count":15,"area":null},{"id":"2412.09925","title":"Simulating Hard Attention Using Soft Attention","created":"2024-12-13","cited_by_count":15,"area":null},{"id":"2307.01201","title":"Schema-learning and rebinding as mechanisms of in-context learning and emergence","created":"2023-06-16","cited_by_count":14,"area":null},{"id":"2401.12947","title":"Transformer-Based Models Are Not Yet Perfect At Learning to Emulate Structural Recursion","created":"2024-01-23","cited_by_count":14,"area":null},{"id":"2408.04532","title":"How Transformers Utilize Multi-Head Attention in In-Context Learning? A Case Study on Sparse Linear Regression","created":"2024-08-08","cited_by_count":14,"area":null},{"id":"2507.02550","title":"Position: A Theory of Deep Learning Must Include Compositional Sparsity","created":"2025-07-03","cited_by_count":14,"area":null},{"id":"2305.03796","title":"Transformer Working Memory Enables Regular Language Reasoning and Natural Language Length Extrapolation","created":"2023-05-05","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":"2402.05785","title":"Limits of Transformer Language Models on Learning to Compose Algorithms","created":"2024-02-08","cited_by_count":13,"area":null},{"id":"2507.12549","title":"The Serial Scaling Hypothesis","created":"2025-07-16","cited_by_count":13,"area":null},{"id":"2410.13502","title":"MathGAP: Out-of-Distribution Evaluation on Problems with Arbitrarily Complex Proofs","created":"2024-10-17","cited_by_count":12,"area":null},{"id":"2412.18040","title":"Theoretical Constraints on the Expressive Power of $\\mathsf{RoPE}$-based Tensor Attention Transformers","created":"2024-12-23","cited_by_count":12,"area":null},{"id":"2305.05364","title":"Large Language Model Programs","created":"2023-05-09","cited_by_count":11,"area":null},{"id":"2503.01805","title":"Depth-Width tradeoffs in Algorithmic Reasoning of Graph Tasks with Transformers","created":"2025-03-03","cited_by_count":11,"area":null},{"id":"2503.19173","title":"Graph neural networks extrapolate out-of-distribution for shortest paths","created":"2025-03-24","cited_by_count":11,"area":null},{"id":"2406.04089","title":"On Limitation of Transformer for Learning HMMs","created":"2024-06-06","cited_by_count":10,"area":null},{"id":"2411.01992","title":"Ask, and it shall be given: On the Turing completeness of prompting","created":"2024-11-04","cited_by_count":10,"area":null},{"id":"2501.04299","title":"Circuit Complexity Bounds for Visual Autoregressive Model","created":"2025-01-08","cited_by_count":10,"area":null},{"id":"2502.16490","title":"On Computational Limits of FlowAR Models: Expressivity and Efficiency","created":"2025-02-23","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":"2410.09605","title":"Training Dynamics of Transformers to Recognize Word Co-occurrence via Gradient Flow Analysis","created":"2024-10-12","cited_by_count":9,"area":null},{"id":"2412.05562","title":"Modern Hopfield Networks Require Chain-of-Thought to Solve $\\mathsf{NC}^1$-Hard Problems","created":"2024-12-07","cited_by_count":9,"area":null},{"id":"2511.17388","title":"Selective Rotary Position Embedding","created":"2025-11-21","cited_by_count":9,"area":null},{"id":"2603.01959","title":"The Expressive Limits of Diagonal SSMs for State-Tracking","created":"2026-03-02","cited_by_count":9,"area":null},{"id":"2309.06054","title":"Breaking through the learning plateaus of in-context learning in Transformer","created":"2023-09-12","cited_by_count":8,"area":null},{"id":"2502.16792","title":"The Role of Sparsity for Length Generalization in Transformers","created":"2025-02-24","cited_by_count":8,"area":null},{"id":"2505.21024","title":"Pause Tokens Strictly Increase the Expressivity of Constant-Depth Transformers","created":"2025-05-27","cited_by_count":8,"area":null},{"id":"2406.02585","title":"Contextual Counting: A Mechanistic Study of Transformers on a Quantitative Task","created":"2024-05-30","cited_by_count":7,"area":null},{"id":"2602.02385","title":"Transformers learn factored representations","created":"2026-02-02","cited_by_count":7,"area":null},{"id":"2607.15178","title":"T^2MLR: Transformer with Temporal Middle-Layer Recurrence","created":"2026-07-16","cited_by_count":7,"area":null},{"id":"2406.17837","title":"Transformer Normalisation Layers and the Independence of Semantic Subspaces","created":"2024-06-25","cited_by_count":6,"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":"2407.21046","title":"Promises and Pitfalls of Generative Masked Language Modeling: Theoretical Framework and Practical Guidelines","created":"2024-07-22","cited_by_count":6,"area":null}],"next_cursor":"100"}