BioKaLMA

BioKaLMA数据集是由南洋理工大学S-Lab创建的，专门用于评估知识感知语言模型归属任务。该数据集包含1085个数据条目，每个条目包括一个问题和回答该问题所需的最小知识集。数据集通过进化问题生成策略构建，旨在控制问题的复杂度和所需知识的范围。BioKaLMA数据集的应用领域主要集中在提高语言模型在处理复杂问题时的准确性和可靠性，特别是在需要精确和事实知识的领域，如金融、法律和医疗治疗。

The BioKaLMA dataset was developed by S-Lab at Nanyang Technological University, and is specifically designed for evaluating the knowledge attribution task of knowledge-aware language models. It contains 1,085 data entries, each including a question and the minimal knowledge set required to answer that question. The dataset is constructed via an evolutionary question generation strategy, aiming to control the complexity of questions and the scope of required knowledge. The primary application scenarios of the BioKaLMA dataset focus on improving the accuracy and reliability of language models when handling complex problems, especially in domains that demand precise and factual knowledge, such as finance, law, and medical treatment.