Papers by Brandon Reagen

1 papers
Spectral Scaling Laws in Language Models: emphHow Effectively Do Feed-Forward Networks Use Their Latent Space? (2025.emnlp-main)

Copied to clipboard

Challenge: Existing scaling laws relate model size to loss, yet overlook how components exploit their latent space.
Approach: They propose to reframe model width selection as a spectral utilization optimization problem . they quantify how many latent directions are meaningfully activated across LLaMA, GPT-2, and nGPT families .
Outcome: The proposed model maximizes the capacity of feed-forward networks by recasting the problem as a spectral utilization optimization problem.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations