Ryo Karakida

4

Papers

1

Total Citations

Papers (4)

Infinite-Width Limit of a Single Attention Layer: Analysis via Tensor Programs

NeurIPS 2025arXiv

Local Loss Optimization in the Infinite Width: Stable Parameterization of Predictive Coding Networks and Target Propagation

Understanding MLP-Mixer as a wide and sparse MLP

Self-attention Networks Localize When QK-eigenspectrum Concentrates