2025-1-2 AI热点速览(补更前1天内容)

文章预览

今日AI领域最新热点新闻: 技术研究热点： 论文: Astraios: Parameter-Efficient Instruction Tuning Code Large Language Models - 提出了参数高效的大语言模型指令微调方法。 论文: Learning to Decode Collaboratively with Multiple Language Models - 探索了多个语言模型协同解码的方法。 论文: Trustworthy Alignment of Retrieval-Augmented Large Language Models via Reinforcement Learning - 提出一种基于强化学习的检索增强大语言模型可信对齐算法。 论文: Unified Hallucination Detection for Multimodal Large Language Models - 解决多模态大语言模型的幻觉问题。 论文: Rethinking Memory and Communication Costs for Efficient Large Language Model Training - 提出新策略PaRO部分冗余优化器，提高大语言模型高效数据并行训练的速度。 论文: Self-Consistency Improves Chain of Thought Reasoning in Language Models - 通过自一致性改进大语言模型的链式思维 ………………………………