Chatbot_instruction_prompts

该数据集是从以下数据生成的:
达图实验室/羊驼
Dahoas/指令-人类-助手-提示
艾伦奈/亲社会对话
数据集已经清理了虚假的条目和工件。它包含约500k的提示和预期的结果。该数据库旨在训练一个指令型模型

This dataset is generated from the following data sources:
Datu Lab/Alpaca
Dahoas/Instruction-Human-Assistant Prompts
AllenAI/Prosocial Dialogues
This dataset has been cleaned of false entries and artifacts. It contains approximately 500k prompts and their corresponding expected outputs. This dataset is intended for training instruction-tuned models.

搜集汇总

背景与挑战

背景概述

Chatbot_instruction_prompts数据集基于Dahoas/instruction-human-assistant-prompts和AllenAI/prosocial-dialog等来源生成，经过清理后包含约50万条提示和预期结果。该数据集旨在用于训练指令型模型，以提升模型在指令跟随方面的性能。