Copertina

Manifold Trajectories in Next-Token Prediction: From Replicator Dynamics to Softmax Equilibrium

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autore principale:	Lee-Jenkins, Christopher R.
Natura:	Preprint
Pubblicazione:	2025
Soggetti:	Machine Learning Artificial Intelligence Dynamical Systems
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

A Minimal Bifurcation Model of Load Imbalance in a Softmax Mixture-of-Experts Router
di: Kiselev, O. M.
Pubblicazione: (2026)

Detecting Invariant Manifolds in ReLU-Based RNNs
di: Eisenmann, Lukas, et al.
Pubblicazione: (2025)

Planning Neural Dynamics with Lie Group Embedding through Supervised Projective Manifold Learning
di: Wang, Tianwei, et al.
Pubblicazione: (2026)

Identifying Equivalent Training Dynamics
di: Redman, William T., et al.
Pubblicazione: (2023)

Out-of-Domain Generalization in Dynamical Systems Reconstruction
di: Göring, Niclas, et al.
Pubblicazione: (2024)

Optimal Recurrent Network Topologies for Dynamical Systems Reconstruction
di: Hemmer, Christoph Jürgen, et al.
Pubblicazione: (2024)

True Zero-Shot Inference of Dynamical Systems Preserving Long-Term Statistics
di: Hemmer, Christoph Jürgen, et al.
Pubblicazione: (2025)

Inverse Depth Scaling From Most Layers Being Similar
di: Liu, Yizhou, et al.
Pubblicazione: (2026)

Dynamical Systems Theory Behind a Hierarchical Reasoning Model
di: Es'kin, Vasiliy A., et al.
Pubblicazione: (2026)

Bayesian Inference in Recurrent Explicit Duration Switching Linear Dynamical Systems
di: Słupiński, Mikołaj, et al.
Pubblicazione: (2024)

Large-Step Training Dynamics of a Two-Factor Linear Transformer Model
di: Balasubramanian, Krishnakumar
Pubblicazione: (2026)

Position: Why a Dynamical Systems Perspective is Needed to Advance Time Series Modeling
di: Durstewitz, Daniel, et al.
Pubblicazione: (2026)

A Hybrid Approach of Transfer Learning and Physics-Informed Modeling: Improving Dissolved Oxygen Concentration Prediction in an Industrial Wastewater Treatment Plant
di: Koksal, Ece S., et al.
Pubblicazione: (2024)

Hybrid Energy-Based Models for Physical AI: Provably Stable Identification of Port-Hamiltonian Dynamics
di: Betteti, Simone, et al.
Pubblicazione: (2026)

Kernel Dynamics under Path Entropy Maximization
di: Das, Jnaneshwar
Pubblicazione: (2026)

Learning Chaotic Systems and Long-Term Predictions with Neural Jump ODEs
di: Krach, Florian, et al.
Pubblicazione: (2024)

Equilibrium Propagation for Non-Conservative Systems
di: Scurria, Antonino Emanuele, et al.
Pubblicazione: (2026)

Horizon-Constrained Rashomon Sets for Chaotic Forecasting
di: Kale, Gauri, et al.
Pubblicazione: (2026)

A-PINN: Auxiliary Physics-informed Neural Networks for Structural Vibration Analysis in Continuous Euler-Bernoulli Beam
di: Saini, Shivani, et al.
Pubblicazione: (2025)

On the dimension of pullback attractors in recurrent neural networks
di: Fadera, Muhammed
Pubblicazione: (2025)

Confabulation dynamics in a reservoir computer: Filling in the gaps with untrained attractors
di: O'Hagan, Jack, et al.
Pubblicazione: (2025)

Sparse identification of nonlinear dynamics and Koopman operators with Shallow Recurrent Decoder Networks
di: Gao, Mars Liyao, et al.
Pubblicazione: (2025)

Language models as master equation solvers
di: Liu, Chuanbo, et al.
Pubblicazione: (2023)

Decomposing heterogeneous dynamical systems with graph neural networks
di: Allier, Cédric, et al.
Pubblicazione: (2024)

A Mathematical Conflict Framework for Contextual Data Modulation
di: Kartal, Hakan Emre
Pubblicazione: (2026)

Improved deep learning of chaotic dynamical systems with multistep penalty losses
di: Chakraborty, Dibyajyoti, et al.
Pubblicazione: (2024)

On the Limitations of Fractal Dimension as a Measure of Generalization
di: Tan, Charlie B., et al.
Pubblicazione: (2024)

The Recurrent Sticky Hierarchical Dirichlet Process Hidden Markov Model
di: Słupiński, Mikołaj, et al.
Pubblicazione: (2024)

Late Fusion Neural Operators for Extrapolation Across Parameter Space in Partial Differential Equations
di: van Tegelen, Eva, et al.
Pubblicazione: (2026)

Inverse Approximation Theory for Nonlinear Recurrent Neural Networks
di: Wang, Shida, et al.
Pubblicazione: (2023)

Memory in Plain Sight: Surveying the Uncanny Resemblances of Associative Memories and Diffusion Models
di: Hoover, Benjamin, et al.
Pubblicazione: (2023)

KPFlow: An Operator Perspective on Dynamic Collapse Under Gradient Descent Training of Recurrent Networks
di: Hazelden, James, et al.
Pubblicazione: (2025)

How to Tame Your LLM: Semantic Collapse in Continuous Systems
di: Wyss, C. M.
Pubblicazione: (2025)

A Hybrid Adaptive Velocity Aided Navigation Filter with Application to INS/DVL Fusion
di: Or, Barak, et al.
Pubblicazione: (2022)

Causal Structure Learning for Dynamical Systems with Theoretical Score Analysis
di: Tagliapietra, Nicholas, et al.
Pubblicazione: (2025)

Learning Interpretable Hierarchical Dynamical Systems Models from Time Series Data
di: Brenner, Manuel, et al.
Pubblicazione: (2024)

Almost-Linear RNNs Yield Highly Interpretable Symbolic Codes in Dynamical Systems Reconstruction
di: Brenner, Manuel, et al.
Pubblicazione: (2024)

When is a System Discoverable from Data? Discovery Requires Chaos
di: Shumaylov, Zakhar, et al.
Pubblicazione: (2025)

Cross-Modal Diffusion for Biomechanical Dynamical Systems Through Local Manifold Alignment
di: Dey, Sharmita, et al.
Pubblicazione: (2025)

SymFlux: deep symbolic regression of Hamiltonian vector fields
di: Evangelista-Alvarado, M. A., et al.
Pubblicazione: (2025)