five

s-nlp/Mistral-7b-0.3-Instruct-DBpedia-HighlyKnown

收藏
Hugging Face2025-02-24 更新2025-04-12 收录
下载链接:
https://hf-mirror.com/datasets/s-nlp/Mistral-7b-0.3-Instruct-DBpedia-HighlyKnown
下载链接
链接失效反馈
官方服务:
资源简介:
这是一个基于DBpedia数据集构建的问答任务数据集,用于研究论文《How Much Knowledge Can You Pack into a LoRA Adapter without Harming LLM?》。数据集包含问题、答案及其变体、预处理问题、贪婪答案、答案的概率、类别、文本段落、段落长度以及是否包含段落的标记等信息。数据集分为一个完整的split,包含约2.1万个示例。

This is a question-answering dataset based on the DBpedia dataset, used for the research paper How Much Knowledge Can You Pack into a LoRA Adapter without Harming LLM?. The dataset includes information such as questions, answers and their aliases, preprocessed questions, greedy answers, probability of answers, categories, text paragraphs, paragraph lengths, and a boolean indicating the presence of a paragraph. The dataset is split into a full set containing approximately 21,000 examples.
提供机构:
s-nlp
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作