PLOT: Enhancing Preference Learning via Optimal Transport

TL;DR

PLOT introduces a novel token-level preference learning method for LLMs using Optimal Transport, improving alignment with human preferences while maintaining stability and semantic coherence.

Contribution

It formulates preference learning as an Optimal Transport problem, enabling globally informed, stable, and robust fine-tuning of language models.

Findings

PLOT outperforms existing methods in preference alignment tasks.

It maintains fluency and coherence in generated text.

Experiments cover diverse preference categories and subpreferences.

Abstract

Preference learning in Large Language Models (LLMs) has advanced significantly, yet existing methods remain limited by modest performance gains, high computational costs, hyperparameter sensitivity, and insufficient modeling of global token-level relationships. We introduce PLOT, which enhances Preference Learning in fine-tuning-based alignment through a token-level loss derived from Optimal Transport. By formulating preference learning as an Optimal Transport Problem, PLOT aligns model outputs with human preferences while preserving the original distribution of LLMs, ensuring stability and robustness. Furthermore, PLOT leverages token embeddings to capture semantic relationships, enabling globally informed optimization. Experiments across two preference categories - Human Values and Logic & Problem Solving - spanning seven subpreferences demonstrate that PLOT consistently improves alignment performance while maintaining fluency and coherence. These results substantiate optimal transport as a principled methodology for preference learning, establishing a theoretically grounded framework that provides new insights for preference learning of LLMs.