新研究通过嘈杂数据和选择性预测解决大语言模型对齐问题

报道来源 [4]

1. arXiv cs.AI TIER_1 English(EN) · Jialiang Wang, Xianming Liu, Xiong Zhou, Hui Liu, Haoliang Li · 2026-07-07 04:00

   面向含噪声偏好的大型语言模型的无偏对齐

   arXiv:2607.03248v1 Announce Type: cross Abstract: The alignment of large language models with human preferences is commonly achieved through Reinforcement Learning from Human Feedback or Direct Preference Optimization. However, these methods are vulnerable to the significant nois…

2. arXiv cs.AI TIER_1 English(EN) · Gaoxiang Luo, Yifan Wu, Sinian Zhang, Aryan Deshwal, Ju Sun · 2026-07-07 04:00

   语言模型与选择性预测的对齐

   arXiv:2607.03528v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly deployed as critical decision-making components in high-stakes real-world AI systems, rendering LLM reliability a foremost practical concern. In this paper, we focus on enhancing LLM r…

3. arXiv cs.CL TIER_1 English(EN) · Sijin Dong, Hiroyuki Shinnou · 2026-07-07 04:00

   具有可证明对齐保证的大型语言模型中的不确定性感知弃权

   arXiv:2607.04430v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly deployed in question answering (QA) systems, yet they may generate hallucinated or misaligned responses without reliable confidence estimates. Uncertainty quantification (UQ) offers a na…

4. arXiv cs.CL TIER_1 English(EN) · Hiroyuki Shinnou · 2026-07-05 17:50

   具有可证明对齐保证的大型语言模型中的不确定性感知弃权

   Large language models (LLMs) are increasingly deployed in question answering (QA) systems, yet they may generate hallucinated or misaligned responses without reliable confidence estimates. Uncertainty quantification (UQ) offers a natural basis for selective answering, where a sys…