Dynamical phase selection controls compute scaling in looped transformers
cond-mat.dis-nn, cond-mat.stat-mech, cs.LG
Submitted: 2026-08-27
Updated: 2026-08-27
Terminology
Sources
- Universal Transformers
- Reasoning with Latent Thoughts: On the Power of Looped Transformers
- Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach
- LoopFormer: Elastic-Depth Looped Transformers for Latent Reasoning via Shortcut Modulation
- Fixed-Point Reasoners: Stable and Adaptive Deep Looped Transformers
- Parcae: Scaling Laws For Stable Looped Language Models
- LayerNorm as Implicit Gain Control in Looped Transformers
- Stability and Generalization in Looped Transformers
- Charting and navigating the space of solutions for recurrent neural networks
- Looped Transformers with Layer Normalization Provably Learn the Power Method
Related papers
- Few-Shot Neuromorphic Vision in a Nonlinear Photonic Network Laser
- Hyperbolic lattices with mass disorder: Phases and phase transitions
- Adaptive Neural Quantum States: A Recurrent Neural Network Perspective
- Machine learning Majorana topology using unsupervised and supervised learning
- Quenched fluctuation-induced force arising from polarization disorder