Temperature sampling is a conventional approach to diversify large language model predictions. As temperature increases, the prediction becomes diverse but also vulnerable to hallucinations -- generating tokens that are sensible but not factual. One common approach to mitigate hallucinations is to provide source/grounding documents and the model is trained to produce predictions that bind to and are attributable to the provided source. It appears that there is a trade-off between diversity and attribution. To mitigate any such trade-off, we propose to relax the constraint of having a fixed temperature over decoding steps, and a mechanism to guide the dynamic temperature according to its relevance to the source through KL-divergence. Our experiments justifies the trade-off, and shows that our sampling algorithm outperforms the conventional top-k and top-p algorithms in conversational question-answering and summarization tasks.
Teacher-Guided Policy Optimization for On-Policy Reasoning Distillation under Large Policy Divergence
School of Computer Science and Engineering, Northeastern University, China(东北大学计算机科学与工程学院);Tsinghua University(清华大学);Meituan(美团);Meta AI;NiuTrans Research, Shenyang, China(新译研究院,沈阳,中国)
Ten Words Only Still Help: Improving Black-Box AI-Generated Text Detection via Proxy-Guided Efficient Re-Sampling
Key Lab of Intelligent Information Processing of Chinese Academy of Sciences(中国科学院智能信息处理重点实验室);Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所);University of Chinese Academy of Sciences(中国科学院大学);Xi’an Jiaotong University(西安交通大学)
$f$-Divergence Regularized RLHF: Two Tales of Sampling and Unified Analyses
Department of Electrical and Computer Engineering, University of Virginia, Virginia, United States(弗吉尼亚大学电气与计算机工程系);Princeton Language and Intelligence, Princeton University, New Jersey, United States(普林斯顿大学语言与智能)
Small RL Controller, Large Language Model: RL-Guided Adaptive Sampling for Test-Time Scaling
University of North Carolina at Chapel Hill(北卡罗来纳大学教堂山分校);University of Maryland College Park(马里兰大学学院市分校);Washington University in St. Louis(圣路易斯华盛顿大学)
Mitigating Tail Narrowing in LLM Self-Improvement via Socratic-Guided Sampling
School of Computer Science, Fudan University(复旦大学计算机科学学院);Meituan(美团);Institute of Modern Languages and Linguistics, Fudan University(复旦大学现代语言与语言学研究所);Key Laboratory of Intelligent Information Processing, Fudan University(复旦大学智能信息处理重点实验室);Macau University of Science and Technology(澳门科技大学)
From Atoms to Chains: Divergence-Guided Reasoning Curriculum for Unlabeled LLM Domain Adaptation
Beijing Key Laboratory of Intelligent Information Technology, School of Computer Science and Technology, Beijing Institute of Technology(北京智能信息科技重点实验室,计算机科学与技术学院,北京理工大学);ByteDance China(字节跳动中国)
Transferable Direct Prompt Injection via Activation-Guided MCMC Sampling
School of Software Engineering, Huazhong University of Science and Technology(华中科技大学软件学院);School of Cyber Science and Engineering, Huazhong University of Science and Technology(华中科技大学网络安全学院);College of Computing and Data Science, Nanyang Technological University(南洋理工大学计算与数据科学学院);Faculty of Data Science, City University of Macau(澳门城市大学数据科学学院)
Enhancing Adversarial Transferability by Balancing Exploration and Exploitation with Gradient-Guided Sampling
School of Computer Science & Software Engineering, Shenzhen University, China(深圳大学计算机科学与软件工程学院);Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ), Shenzhen, China(广东省人工智能与数字经济发展实验室(深圳));Guangdong Provincial Key Laboratory of Intelligent Information Processing, Shenzhen University, China(广东省智能信息处理省级重点实验室);School of Computer Science, University of Exeter, U.K.(埃克塞特大学计算机科学学院);Department of Computing and Information Technology, Great Bay University, China(大鹏大学计算与信息技术系);Computer Vision Institute, School of Artificial Intelligence, Shenzhen University, China(人工智能学院计算机视觉研究所)
Diversifying Neural Text Generation with Part-of-Speech Guided Softmax and Sampling
Wangxuan Institute of Computer Technology, Peking University(北京大学王选计算机研究所);Center for Data Science, Peking University(北京大学数据科学中心);The MOE Key Laboratory of Computational Linguistics, Peking University(北京大学计算语言学教育部重点实验室)
Deep Generative Sampling in the Dual Divergence Space: A Data-efficient & Interpretative Approach for Generative AI
Morgan Stanley(摩根士丹利);NIST(美国国家标准与技术研究院);Georgetown University(乔治城大学);IBM Research(IBM研究院);Mila - Quebec AI Institute(米拉-魁北克人工智能研究所);Université de Montréal(蒙特利尔大学)