Lots-of-LoRAs/task138_detoxifying-lms_classification_fluency

该数据集名为Natural Instructions，主要用于文本生成任务。数据集包含312个训练样本、39个验证样本和39个测试样本。每个样本包含输入、输出和ID三个特征。数据集的创建是通过众包完成的，语言为英语，许可证为Apache-2.0。数据集的相关论文包括《Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP Tasks》和《Compress then Serve: Serving Thousands of LoRA Adapters with Little Overhead》。

The dataset is named Natural Instructions and is primarily used for text generation tasks. It contains 312 training samples, 39 validation samples, and 39 test samples. Each sample includes three features: input, output, and ID. The dataset was created through crowdsourcing, is in English, and is licensed under Apache-2.0. Relevant papers for the dataset include Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+ NLP Tasks and Compress then Serve: Serving Thousands of LoRA Adapters with Little Overhead.

原始信息汇总

数据集概述

基本信息

• 数据集名称: task138_detoxifying-lms_classification_fluency
• 任务类别: 文本生成
• 语言: 英语
• 许可证: Apache 2.0
• 数据集创建者: 众包
• 注释创建者: 众包

数据集结构

• 配置名称: plain_text
• 特征:
  • input: 字符串类型
  • output: 字符串类型
  • id: 字符串类型
• 数据分割:
  • 训练集: 312个样本
  • 验证集: 39个样本
  • 测试集: 39个样本