Research: Steering LLMs for Localized Cultural Content
simi_97k · x · 2026-07-12
Introduces the upcoming paper "Steering LLMs for Culturally Localized Generation" to be published at COLM 2026. The study explores which culture LLMs default to aligning with in open-ended questions and how to change this tendency. Methodologically, the research extracts a small number of cultural assertions (around 200) from public datasets, identifies sparse autoencoder features with high mutual information for <sentence, country> pairs, and packages them into Cultural Embeddings (CuE).
Related event: Steering LLMs for Culturally Localized Generation(5 posts)→
More from Research
- TRL adds Score Centering, a simple fix for RLHF training instability — lvwerra · 2026-10-27
- Latent-Foresight: world models that learn to predict their own future — abursuc · 2026-10-27
- UK AISI papers: no good safeguards yet against misuse of biological AI tools — davidmanheim · 2026-10-27
- CMU's XYBench: LLMs answer queries literally and reinforce user misconceptions — valentina__py · 2026-10-27
- arXiv advice from T. Dietterich: wrong primary category will almost surely get your paper on hold — tdietterich · 2026-10-27
- arXiv has 4,000+ papers on hold amid submission surge; moderator shares how to avoid it — tdietterich · 2026-10-27