Recently, Large Language Models (LLMs) have demonstrated outstanding performance across a wide range of downstream language tasks. Temperature sampling is a commonly used decoding strategy for LLMs' generation process. However, a fixed temperature parameter is used in most cases, which may not always be an optimal choice for balancing generation quality and diversity. In this paper, we propose an effective Entropy-based Dynamic Temperature (EDT) Sampling method, to achieve a more balanced performance in terms of both generation quality and diversity by dynamically selecting the temperature parameter. Additionally, we also show model performance and comprehensive analyses for 4 different generation benchmarks. Our experiments show that EDT significantly outperforms the existing strategies across different tasks.
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)
PANDAS: Improving Many-shot Jailbreaking via Positive Affirmation, Negative Demonstration, and Adaptive Sampling
University of Toronto, Vector Institute(多伦多大学,向量研究所);University of Oxford(牛津大学);Polytechnique Montréal, Mila - Quebec AI Institute, University of Toronto(蒙特利尔理工学院,魁北克人工智能研究所,多伦多大学)
Improving Sampling Methods for Fine-tuning SentenceBERT in Text Streams
Instituto Federal de Santa Catarina (IFSC)(圣卡塔琳娜联邦学院(IFSC));Pontifícia Universidade Católica do Paraná (PUCPR)(巴拉那天主教大学(PUCPR));École de Technologie Supérieure (ÉTS), Université du Québec(魁北克大学高等技术学院(ÉTS));Universidade Estadual de Ponta Grossa (UEPG)(庞塔格罗萨州立大学(UEPG))
Improving Word Representations: A Sub-sampled Unigram Distribution for Negative Sampling
The Chinese University of Hong Kong(香港中文大学)
Word 2 Vec是最流行的词表示模型,在文献中得到了广泛的研究。然而,其负采样噪声分布是由经验试验决定的,其最优性一直被忽视。我们认为,该分布是一个次优的选择,并建议使用子采样的unigram分布更好的负采样。我们的贡献包括:(1)提出了语义量化的概念,并推导出了适合不同训练语料的次采样率;(2)通过大量的评价任务,证明了该方法在负采样和噪声对比估计方面的优势;(3)提出了一种语义加权模型,用于MSR句子完成任务,结果有了很大的改进。我们的工作不仅提高了词向量的质量,而且有利于当前对Word 2 Vec的理解。
Improving Simultaneous Machine Translation with Monolingual Data
JD Explore Academy(京东探索研究院)
同步机器翻译(SiMT)通常通过从整句神经机器翻译(NMT)模型中提取序列级知识(Seq-KD)来完成。然而,NMT和SiMT之间仍然存在显著的性能差距。在这项工作中,我们建议利用单语数据来改进SiMT,该SiMT学生将Seq-KD提取的双语数据和外部单语数据相结合。在En-Zh和En-Ja新闻领域语料库上的初步实验表明,单语数据可以显着提高翻译质量(例如,+3.15 BLEU on En-Zh)。受同声传译人员行为的启发,我们提出了一种新的SiMT单语采样策略,同时考虑块长度和单调性。实验结果表明,该采样策略避免了SiMT的关键问题--幻觉,性能优于随机采样策略(以及其他传统的典型NMT单语采样策略),并具有更好的可扩展性。我们在En-Zh和En-Ja上实现了平均+0.72 BLEU改进。数据和代码可以在https://github.com/hexuandeng/Mono4SiMT上找到。
Improving Sampling for Masked Diffusion Models via Information Gain
Department of Computer Science and Technology, Tsinghua University(清华大学计算机科学与技术系);Massachusetts Institute of Technology(麻省理工学院);Shanghai Jiao Tong University(上海交通大学);Beihang University(北航);College of AI, Tsinghua University(清华大学人工智能学院)