Uni-DPO: A Unified Paradigm for Dynamic Preference Optimization of LLMs
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916041780625408 |
|---|---|
| author | Peng, Shangpin Wang, Weinong Tian, Zhuotao Yang, Senqiao Wu, Xing Xu, Haotian Zhang, Chengquan Isobe, Takashi Hu, Baotian Zhang, Min |
| author_facet | Peng, Shangpin Wang, Weinong Tian, Zhuotao Yang, Senqiao Wu, Xing Xu, Haotian Zhang, Chengquan Isobe, Takashi Hu, Baotian Zhang, Min |
| contents | Direct Preference Optimization (DPO) has emerged as a cornerstone of reinforcement learning from human feedback (RLHF) due to its simplicity and efficiency. However, existing DPO-based methods typically treat all preference pairs equally, overlooking substantial variations in data quality and learning difficulty, which leads to inefficient data utilization and suboptimal performance. To address this limitation, we propose Uni-DPO, a unified dynamic preference optimization framework that jointly considers (a) the inherent quality of preference pairs and (b) the model's evolving performance during training. By adaptively reweighting samples based on both factors, Uni-DPO enables more effective use of preference data and achieves superior performance. Extensive experiments across models and benchmarks demonstrate the effectiveness and generalization of Uni-DPO. On textual tasks, Gemma-2-9B-IT fine-tuned with Uni-DPO surpasses the leading LLM, Claude 3 Opus, by 6.7 points on Arena-Hard. On mathematical and multimodal tasks, Uni-DPO consistently outperforms baseline methods across all benchmarks, providing strong empirical evidence of its effectiveness and robustness. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_10054 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Uni-DPO: A Unified Paradigm for Dynamic Preference Optimization of LLMs Peng, Shangpin Wang, Weinong Tian, Zhuotao Yang, Senqiao Wu, Xing Xu, Haotian Zhang, Chengquan Isobe, Takashi Hu, Baotian Zhang, Min Machine Learning Artificial Intelligence Computation and Language Computer Vision and Pattern Recognition Direct Preference Optimization (DPO) has emerged as a cornerstone of reinforcement learning from human feedback (RLHF) due to its simplicity and efficiency. However, existing DPO-based methods typically treat all preference pairs equally, overlooking substantial variations in data quality and learning difficulty, which leads to inefficient data utilization and suboptimal performance. To address this limitation, we propose Uni-DPO, a unified dynamic preference optimization framework that jointly considers (a) the inherent quality of preference pairs and (b) the model's evolving performance during training. By adaptively reweighting samples based on both factors, Uni-DPO enables more effective use of preference data and achieves superior performance. Extensive experiments across models and benchmarks demonstrate the effectiveness and generalization of Uni-DPO. On textual tasks, Gemma-2-9B-IT fine-tuned with Uni-DPO surpasses the leading LLM, Claude 3 Opus, by 6.7 points on Arena-Hard. On mathematical and multimodal tasks, Uni-DPO consistently outperforms baseline methods across all benchmarks, providing strong empirical evidence of its effectiveness and robustness. |
| title | Uni-DPO: A Unified Paradigm for Dynamic Preference Optimization of LLMs |
| topic | Machine Learning Artificial Intelligence Computation and Language Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2506.10054 |