LKF-unlearning_Jewell_new_config_no_filter_rephrasings_final
收藏资源简介:
该数据集名为forget_eval_rephrasings,包含63个样本,数据总量约为128.8KB。其核心结构包括三个主要字段:question(问题)、answer(答案)和label(标签),以及扩展字段:10个以q_gemini为前缀的字段(q_gemini1至q_gemini10)和5个以blank_gemini为前缀的字段(blank_gemini1至blank_gemini5),所有字段均为文本字符串类型。从名称推断,该数据集很可能设计用于评估语言模型(可能与Gemini模型相关)在“遗忘”或“问题重新表述”任务上的性能或行为,label字段可能用于标注任务相关的类别或评估结果。数据集适用于自然语言处理中的模型评估、遗忘学习或文本改写分析等研究场景。
The dataset is named forget_eval_rephrasings and contains 63 samples with a total data size of approximately 128.8KB. Its core structure includes three main fields: question, answer, and label, along with extended fields: 10 fields prefixed with q_gemini (q_gemini1 to q_gemini10) and 5 fields prefixed with blank_gemini (blank_gemini1 to blank_gemini5), all of which are text string types. Based on the name, the dataset is likely designed to evaluate the performance or behavior of language models (possibly related to the Gemini model) on forgetting or question rephrasing tasks, with the label field potentially used to annotate task-related categories or evaluation results. The dataset is suitable for research scenarios such as model evaluation, forgetting learning, or text rephrasing analysis in natural language processing.
数据集概述
基本信息
- 数据集名称: LKF-unlearning_Jewell_new_config_no_filter_rephrasings_final
- 数据集地址: https://huggingface.co/datasets/Bibhabasu/LKF-unlearning_Jewell_new_config_no_filter_rephrasings_final
数据特征
该数据集包含以下字段,所有字段均为字符串类型(string):
- question: 问题文本
- answer: 答案文本
- label: 标签
- q_gemini1 至 q_gemini10: 由 Gemini 模型生成的10个不同版本的问题
- blank_gemini1 至 blank_gemini5: 由 Gemini 模型生成的5个空白版本
数据集划分
- 唯一划分:
forget_eval_rephrasings - 示例数量: 63 个
- 数据集大小: 128,829 字节
- 下载大小: 82,106 字节
配置信息
- 配置名称:
default - 数据文件路径:
data/forget_eval_rephrasings-*




