InCo-DPO: Balancing Distribution Shift and Data Quality for Enhanced Preference Optimization
This work addresses a key bottleneck in preference optimization for AI language models, offering an incremental improvement over existing methods.
The paper tackles the problem of balancing distribution shift and data quality in Direct Preference Optimization (DPO) for language models by proposing InCo-DPO, which integrates on-policy and off-policy data to achieve a state-of-the-art win rate of 60.8 on Arena-Hard.
Direct Preference Optimization (DPO) optimizes language models to align with human preferences. Utilizing on-policy samples, generated directly by the policy model, typically results in better performance due to its distribution consistency with the model compared to off-policy samples. This paper identifies the quality of candidate preference samples as another critical factor. While the quality of on-policy data is inherently constrained by the capabilities of the policy model, off-policy data, which can be derived from diverse sources, offers greater potential for quality despite experiencing distribution shifts. However, current research mostly relies on on-policy data and neglects the value of off-policy data in terms of data quality, due to the challenge posed by distribution shift. In this paper, we propose InCo-DPO, an efficient method for synthesizing preference data by integrating on-policy and off-policy data, allowing dynamic adjustments to balance distribution shifts and data quality, thus finding an optimal trade-off. Consequently, InCo-DPO overcomes the limitations of distribution shifts in off-policy data and the quality constraints of on-policy data. We evaluated InCo-DPO with the Alpaca-Eval 2.0 and Arena-Hard benchmarks. Experimental results demonstrate that our approach not only outperforms both on-policy and off-policy data but also achieves a state-of-the-art win rate of 60.8 on Arena-Hard with the vanilla DPO using Gemma-2 model.