Beyond the Training Horizon: Mechanisms and Limits of Length Generalization in Looped Transformers
cs.LG, cs.AI
Submitted: 2026-09-27
Updated: 2026-09-27
Terminology
Sources
- A Mechanistic Analysis of Looped Reasoning Language Models
- Looped Transformers for Length Generalization
- Bridging the Gap Between Latent and Explicit Reasoning with Looped Transformers
- Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach
- Loop, Think, & Generalize: Implicit Reasoning in Recurrent-Depth Transformers
- Stability and Generalization in Looped Transformers
- Measuring Faithfulness in Chain-of-Thought Reasoning
- Do Latent-CoT Models Think Step-by-Step? A Mechanistic Study on Sequential Reasoning Tasks
- Explicitly Encoding Structural Symmetry is Key to Length Generalization in Arithmetic Tasks
- Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small
- Stabilizing Recurrent Dynamics for Test-Time Scalable Latent Reasoning in Looped Language Models
- Transformers Can Achieve Length Generalization But Not Robustly
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks