The Role of Feed-Forward Layers in Transformer Dynamics

arXiv:2609.36230 · cs.LG, math.DS · Submitted 2026-09-28 · Read on arXiv

cs.LG, math.DS

Submitted: 2026-09-28

Updated: 2026-09-28

Terminology

• Transformer — 1× in this paper · explained in 2 episodes

Sources

• Quantifying Concentration Phenomena of Mean-Field Transformers in the Low-Temperature Regime
• Perceptrons and localization of attention's mean-field landscape
• Emergence of meta-stable clustering in mean-field transformer models
• A multiscale analysis of mean-field transformers in the moderate interaction regime
• A Unified Perspective on the Dynamics of Deep Transformers
• Transformers are Universal In-context Learners
• Dynamic metastability in the self-attention model
• Universal Approximation Theorem for a Single-Layer Transformer
• AlphaFold two years on: validation and impact
• Transformers Meet In-Context Learning: A Universal Approximation Theory
• On the Diverse Dynamical Behaviors Arising in Deep Linear Transformers
• Gemini: A Family of Highly Capable Multimodal Models
• The Llama 3 Herd of Models

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: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks