Intuitive Fine-Tuning: Towards Simplifying Alignment into a Single Process

TL;DR

This paper introduces Intuitive Fine-Tuning (IFT), a unified approach that combines supervised fine-tuning and preference optimization into a single process, enhancing language model alignment with human preferences.

Contribution

The paper presents a novel interpretation of SFT and PO as sub-processes within an MDP, and proposes IFT, which integrates both into one process using a temporal residual connection.

Findings

IFT performs comparably or better than SFT and PO on several tasks.

IFT effectively captures the model's intuitive sense of entire answers.

Experimental results validate IFT's effectiveness in various generation tasks.

Abstract

Supervised Fine-Tuning (SFT) and Preference Optimization (PO) are key processes for aligning Language Models (LMs) with human preferences post pre-training. While SFT excels in efficiency and PO in effectiveness, they are often combined sequentially without integrating their optimization objectives. This approach ignores the opportunities to bridge their paradigm gap and take the strengths from both. In this paper, we interpret SFT and PO with two sub-processes -- Preference Estimation and Transition Optimization -- defined at token level within the Markov Decision Process (MDP). This modeling shows that SFT is only a special case of PO with inferior estimation and optimization. PO estimates the model's preference by its entire generation, while SFT only scores model's subsequent predicted tokens based on prior tokens from ground truth answer. These priors deviates from model's distribution, hindering the preference estimation and transition optimization. Building on this view, we introduce Intuitive Fine-Tuning (IFT) to integrate SFT and PO into a single process. Through a temporal residual connection, IFT brings better estimation and optimization by capturing LMs' intuitive sense of its entire answers. But it solely relies on a single policy and the same volume of non-preference-labeled data as SFT. Our experiments show that IFT performs comparably or even superiorly to SFT and some typical PO methods across several tasks, particularly those require generation, reasoning, and fact-following abilities. An explainable Frozen Lake game further validates the effectiveness of IFT for getting competitive policy.