MixDPO: Modeling Preference Strength for Pluralistic Alignment

TL;DR

MixDPO introduces a novel approach to modeling varying preference strengths in language model alignment, improving performance and capturing heterogeneity in human judgments across diverse datasets.

Contribution

It generalizes Direct Preference Optimization by explicitly modeling preference strength variation, enhancing alignment with heterogeneous human preferences.

Findings

Improves aggregate alignment performance (+11.2 points on Pythia-2.8B)

Preserves subgroup preferences across datasets

Effectively captures preference heterogeneity in training data

Abstract

Preference based alignment objectives implicitly assume that all human preferences are expressed with equal strength. In practice, however, preference strength varies across individuals and contexts -- a phenomenon established in behavioral economics and discrete choice theory. This mismatch limits the ability of existing objectives to faithfully capture heterogeneous human judgments. Inspired by this literature, we introduce Mixed Logit Direct Preference Optimization (MixDPO), a generalization of Direct Preference Optimization that models variation in preference strength. MixDPO enables alignment objectives to capture heterogeneity in how strongly preferences are expressed across training examples. We evaluate MixDPO on three preference datasets using two open-weight language models. Across datasets, MixDPO improves aggregate alignment performance (+11.2 points on Pythia-2.8B) while preserving subgroup level preferences, with the largest gains appearing in settings with higher inferred preference heterogeneity. MixDPO makes preference heterogeneity explicit through learned strength distributions. We release our code for reproducibility.