CuE: Activating and Suppressing Cultural Info in LLMs

simi_97k · x · 2026-07-12

Introduces the composition of Cultural Embeddings (CuE). Using Neuronpedia, the research team analyzed the sparse autoencoder features of the Gemma-2-9B model, discovering that cultural information within the model is shared, activated, and suppressed in an interpretable manner.

Original post →

More from Research

Research channel →