ORPO: Monolithic Preference Optimization without Reference Model (Paper Explained)

Paper: https://arxiv.org/abs/2403.07691

Abstract:
While recent preference alignment algorithms for language models have demonstrated promising results, supervised fine-tuning (SFT) remains imperative for achieving successful convergence. In this paper, we study the crucial role of SFT within the context of preference alignment, emphasizing that a minor penalty for the disfavored generation style is sufficient for preference-aligned SFT. Building on this foundation, we introduce a straightforward and innovative reference model-free monolithic odds ratio preference optimization algorithm, ORPO, eliminating the necessity for an additional …

abstract algorithms alignment building context convergence fine-tuning foundation language language models paper results role sft study style supervised fine-tuning while