Tracking States or Tracking Cosets? An Algebraic Account of Learned State Tracking
cs.LG, cs.AI
Submitted: 2026-09-24
Updated: 2026-09-24
Terminology
Sources
- Hultman numbers, polygon gluings and matrix integrals
- GPT-NeoX-20B: An Open-Source Autoregressive Language Model
- A Toy Model of Universality: Reverse Engineering How Networks Learn Group Operations
- Rethinking State Tracking in Recurrent Models Through Error Control Dynamics
- Neural Networks and the Chomsky Hierarchy
- Learning Dynamics of Chain-of-Thought State Tracking in a Solvable Transformer Model
- Alternating Gradient Flows: A Theory of Feature Learning in Two-layer Neural Networks
- A Held-Out Transition-Pair Falsifier for Long-Horizon Non-Abelian State Tracking
- (How) Do Language Models Track State?
- Transformers Learn Shortcuts to Automata
- Decoupled Weight Decay Regularization
- Sequential Group Composition: A Window into the Mechanics of Deep Learning
- A Little Depth Goes a Long Way: The Expressive Power of Log-Depth Transformers
- The Illusion of State in State-Space Models
- Recirculation
- Progress measures for grokking via mechanistic interpretability
- Show Your Work: Scratchpads for Intermediate Computation with Language Models
- Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets
- Marginals Before Conditionals
- Learning words in groups: fusion algebras, tensor ranks and grokking
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