JoshuaFreeman/raw-fact-extraction

该数据集包含两个配置：joshua_entities_longfact_Meta-Llama-3.1-8B-Instruct 和 longfact_Meta-Llama-3.1-8B-Instruct。每个配置都包括多个特征字段，如数据集名称、子集、原始数据集索引、原始数据集划分、查询ID、完成索引、对话内容和完成内容等。对话内容是一个列表，包含内容和角色。事实字段也是一个列表，包含事实、ID、文本中的索引、标签、标签元数据、句子和跨度等信息。数据集有一个测试划分，并且提供了数据集大小和下载大小。

The dataset consists of two configurations: joshua_entities_longfact_Meta-Llama-3.1-8B-Instruct and longfact_Meta-Llama-3.1-8B-Instruct. Each configuration includes multiple feature fields such as dataset name, subset, original dataset index, original dataset split, query ID, completion index, conversation content, and completion content, etc. The conversation content is a list that includes content and role. The facts field is also a list with details like fact, ID, index in text, label, labeling metadata, sentence, and span. The dataset has a test split, and the dataset size and download size are provided.