Steerable Cultural Preference Optimization of Reward Models

Researchers introduced Steerable Cultural Preference Optimization (SCPO), a reward model training algorithm that balances diverse cultural preferences in large language models. SCPO improved minority reward model performance by up to 7 points across two datasets and seven countries, while being up to 280% more training data-efficient than full-data finetuning. The method also mitigated excessive bias toward any subcommunity.

arXiv:2606.18606v1 Announce Type: new Abstract: It is essential for large language model LLM technology to serve many different cultural sub-communities in a manner that is acceptable to each community. However, research on LLM alignment has so far predominantly focused on predicting a unified response preference of annotators from certain regions. This paper aims to advance the development of alignment models with a more global outlook, that are able to accurately represent the preferences of subcommunities and do not exhibit excessive bias towards any of them. We focus on the development of reward models for this purpose and present a novel reward model training algorithm SCPO that can incorporate diverse cultural preferences in a balanced manner. Our method results in performance increases of the minority reward model of up to 7 points over the baseline model across two datasets, PRISM and GlobalOpinionQA, and across 7 countries. SCPO is up to 280% more training data-efficient than full-data finetuning of reward models. In addition, we perform analysis of bias by separately evaluating on the preference of subcommunities and show that excessive bias is mitigated via our weighting method. Our code is available at https://github.com/minsik-ai/Steerable-Cultural-Preference