大型语言模型已在各类自然语言处理任务上取得令人瞩目的性能。然而,迄今为止,对它们的评估主要集中在输入上下文中所有信息都与解决任务相关的基准上。在本研究中,我们调查大型语言模型的易受干扰性,即模型解决问题的准确率会如何受到无关上下文的影响。具体而言,我们提出了 Grade-School Math with Irrelevant Context(GSM-IC,含无关上下文的小学数学题),这是一个在题目描述中包含无关信息的算术推理数据集。我们利用该基准衡量面向大型语言模型的前沿提示技术的易受干扰性,发现当加入无关信息时,模型性能会显著下降。我们还找出了若干缓解这一缺陷的方法,例如采用 self-consistency(自洽性)解码,以及在提示中加入一条指示,要求语言模型忽略无关信息。
英文摘要:
Large language models have achieved impressive performance on various natural language processing tasks. However, so far they have been evaluated primarily on benchmarks where all information in the input context is relevant for solving the task. In this work, we investigate the distractibility of large language models, i.e., how the model problem-solving accuracy can be influenced by irrelevant context. In particular, we introduce Grade-School Math with Irrelevant Context (GSM-IC), an arithmetic reasoning dataset with irrelevant information in the problem description. We use this benchmark to measure the distractibility of cutting-edge prompting techniques for large language models, and find that the model performance is dramatically decreased when irrelevant information is included. We also identify several approaches for mitigating this deficiency, such as decoding with self-consistency and adding to the prompt an instruction that tells the language model to ignore the irrelevant information.
How Is LLM Reasoning Distracted by Irrelevant Context? An Analysis Using a Controlled Benchmark
University of Arizona(亚利桑那大学);MOE Key Lab of Computational Linguistics, Peking University(北京大学计算语言学国家重点实验室);University of California, Santa Barbara(加州大学圣巴巴拉分校)
Medical large language models are easily distracted
NYU Langone Medical Center(纽约大学朗格尼医学中心);The University of Texas at Austin(德克萨斯大学奥斯汀分校);Washington University School of Medicine in St. Louis(圣路易斯华盛顿大学医学院);New York University(纽约大学)
Negated and Misprimed Probes for Pretrained Language Models: Birds Can Talk, But Cannot Fly
在Petroni等人(2019)的基础上,我们提出了两个新的探测任务,分析存储在预训练语言模型(PLM)中的事实知识。(1)否定。我们发现,PLM没有区分否定型(“Birds cannot [MASK]”)和非否定型(“Birds can [MASK]”)完形填空题。(2)启动错误。受人类心理学中启动方法的启发,我们在完形填空题中加入了“误启动”(“谈话?鸟类可以[面具]”)。我们发现,PLM很容易被误启动分心。这些结果表明,PLM仍然有很长的路要走,充分学习人类一样的事实知识。
Activation-aware Probe-Query: Effective Key-Value Retrieval for Long-Context LLMs Inference
The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州));The Hong Kong University of Science and Technology(香港科技大学);The Hong Kong Polytechnic University(香港理工大学);South China Normal University(华南师范大学);University College London(伦敦大学学院)
CUB: Benchmarking Context Utilisation Techniques for Language Models
Chalmers University of Technology(查尔姆斯理工大学);University of Gothenburg(哥德堡大学);Seoul National University(首尔国立大学);University of Copenhagen(哥本哈根大学);Ewha Womans University(成均馆大学)