The Model Shape Behind Scaling Laws: A Gradient Superposition Perspective
cs.LG
Submitted: 2026-05-27
Updated: 2026-10-03
Terminology
Sources
- Feature learning as alignment: a structural property of gradient descent in non-linear neural networks
- Universal Transformers
- Don't be lazy: CompleteP enables compute-efficient deep transformers
- Toy Models of Superposition
- Scaling and evaluating sparse autoencoders
- Surprises in High-Dimensional Ridgeless Least Squares Interpolation
- Scaling Laws for Autoregressive Generative Modeling
- Deep Learning Scaling is Predictable, Empirically
- Training Compute-Optimal Large Language Models
- Scaling Laws for Neural Language Models
- The Depth-to-Width Interplay in Self-Attention
- Revisiting the Shape Convention of Transformer Language Models
- Neural Scaling Universality: If Exponents Are Fixed, Time to Understand Coefficients
- Superposition Yields Robust Neural Scaling
- Gemstones: A Model Suite for Multi-Faceted Scaling Laws
- Sensitivity and Generalization in Neural Networks: an Empirical Study
- The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scale
- Mechanism of feature learning in deep fully connected networks and kernel machines that recursively learn features
- Jumping Ahead: Improving Reconstruction Fidelity with JumpReLU Sparse Autoencoders
- A Constructive Prediction of the Generalization Error Across Scales
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