conftrace_

infinite width dynamics

2 papers

Explore in graph

Co-occurring keywords · papers mentioning both

training-test loss gap (1) rich regime (1) compute optimal (1) task similarity (1) random feature model (1) lazy regime (1) scaling law (1) gradient descent (1) feature learning (1) catastrophic forgetting (1)

Papers

The Importance of Being Lazy: Scaling Limits of Continual Learning ICML 2025 A Dynamical Model of Neural Scaling Laws ICML 2024