研究人员在直接偏好优化（DPO）方面取得了几项进展，DPO是一种用于将大型语言模型（LLM）与人类偏好对齐的方法。AdaDPO引入了自适应系数来平衡梯度更新，提高了效率并减轻了长度偏差，在基准测试中表现优于标准DPO。Uni-DPO提供了一个统一的动态框架，根据数据质量和模型性能自适应地重新加权样本，在各种任务上取得了优于Claude 3 Opus的卓越结果。此外，AttentionPO利用LLM自身的注意力机制来加权token，使其具有内容感知能力，并提高了在基准测试中的性能。
AI
arXiv:2605.28440v1 Announce Type: new Abstract: DPO has become a widely adopted alternative to RLHF for aligning LLMs with human preferences, eliminating the need for a separate reward model or RL loop. Recent theoretical analysis uncovers an asymmetric gradient behavior in DPO: …
DPO has become a widely adopted alternative to RLHF for aligning LLMs with human preferences, eliminating the need for a separate reward model or RL loop. Recent theoretical analysis uncovers an asymmetric gradient behavior in DPO: the loss suppresses dispreferred responses subst…
arXiv:2506.10054v4 Announce Type: replace-cross Abstract: 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 preferenc…
arXiv:2509.10515v1 Announce Type: cross Abstract: Offline preference optimization methods are efficient for large language models (LLMs) alignment. Direct Preference optimization (DPO)-like learning, one of the most popular approaches, stands out for its efficiency in reward mode…
arXiv:2605.21883v1 Announce Type: new Abstract: Direct Preference Optimization (DPO) aligns Large Language Models with human preferences without the need for a separate reward model. However, DPO treats all tokens in responses equally, neglecting the differing importance of indiv…
Direct Preference Optimization (DPO) aligns Large Language Models with human preferences without the need for a separate reward model. However, DPO treats all tokens in responses equally, neglecting the differing importance of individual tokens. Existing token-level PO methods co…
Together AI now supports DPO fine-tuning. Learn how Direct Preference Optimization aligns language models with human preferences — with code examples and technical details.
