方法证据记录
Direct Preference Optimization
Direct Preference Optimization (DPO) is a training method introduced by Rafailov et al. in 2023 that aligns language models with human preferences without requiring an explicit reward model. By directly optimizing for preference pairs (better response vs worse response), DPO simplifies the training pipeline compared to reinforcement learning from human feedback (RLHF).
源记录
引文逐字复制自方法源记录。这些引文不代表任何层级的验证。
Direct Preference Optimization: Your Language Model is Secretly a Reward Model
分类方法记录 · ml-model / deep-learning
打开完整方法 精选声明
声明已持久化到证据分类账中，每个声明都有自己的评估。
尚无精选声明
当分类账中没有声明时，此视图不会自行创建声明评估。
相关方法
从方法图中生成，显示为机器建议的关系 — 不推断任何证据声明。