SGDPO: Self-Guided Direct Preference Optimization for Language Model Alignment
This work addresses the problem of more resilient and effective language model alignment for AI applications, representing an incremental improvement over existing methods.
The paper tackles the limitations of Direct Preference Optimization (DPO) in aligning Large Language Models with human values by proposing SGDPO, a self-guided algorithm that improves performance by up to 9.19% on benchmarks.
Direct Preference Optimization (DPO) is broadly utilized for aligning Large Language Models (LLMs) with human values because of its flexibility. Despite its effectiveness, it has been observed that the capability of DPO to generate human-preferred response is limited and the results of DPO are far from resilient. To address these limitations, in this paper we propose a novel Self-Guided Direct Preference Optimization algorithm, i.e., SGDPO, which incorporates a pilot term to steer the gradient flow during the optimization process, allowing for fine-grained control over the updates of chosen and rejected rewards. We provide a detailed theoretical analysis of our proposed method and elucidate its operational mechanism. Furthermore, we conduct comprehensive experiments on various models and benchmarks. The extensive experimental results demonstrate the consistency between the empirical results and our theoretical analysis and confirm the effectiveness of our proposed approach (up to 9.19% higher score).