arXiv · cs.AI · Austin Tsai, Saurav Nanda, Ganapathy Parthasarathy, Bhuvnesh Kumar, Parivesh Choudhary · 6 小时前模型评分41
独立复现直接纳什优化（DNO）：Llama 3.1 8B 与 Mistral 7B 的对齐实证研究
Aligning Language Models with General Preferences: An Empirical Study of Direct Nash Optimization
导读
研究者独立复现了直接纳什优化（DNO，Direct Nash Optimization）的实用版本 DNO-Prct，在 Llama 3.1 8B Instruct 和 Mistral 7B Instruct v0.1 上验证其对齐效果。
分类置信度
中 · 待人工复核
- 只有单一媒体来源，需要人工复核。
来源级别：媒体与个人
事件类型：研究发布
证据摘录
- “We re-implement DNO-Prct independently and study it on two open-weight instruct models, Llama 3.1 8B Instruct and Mistral 7B Instruct v0.1, with GPT-4o or a larger open-weight model of the same family as the annotator.”
- “All results are single runs, and GPT-4o is both the teacher and the evaluation judge; we discuss what this implies for the reported gains.”
- “Experiments were run in 2024 with the models and APIs available then”
来源：arXiv · cs.AI · arxiv.org