Skip to content
Sagan

Paper

Asymmetric Scaling Laws from Sparse Features

Unreadunread

AI summary

arXiv:2605. 23591v1 Announce Type: new Abstract: We introduce a model for neural scaling laws under sparse activations.