DailyDialog

DailyDialog 是一个高质量的多轮开放域英语对话数据集。它包含 13,118 个对话，分为包含 11,118 个对话的训练集和每个包含 1000 个对话的验证集和测试集。平均而言，每个对话大约有 8 个发言者轮次，每轮大约 15 个标记。

来源：http://yanran.li/dailydialog

DailyDialog is a high-quality multi-turn open-domain English dialogue dataset. It contains 13,118 dialogues, which are divided into a training set with 11,118 dialogues, as well as a validation set and a test set each containing 1000 dialogues. On average, each dialogue has approximately 8 speaker turns, with each turn consisting of around 15 tokens. Source: http://yanran.li/dailydialog

搜集汇总

背景与挑战

背景概述

DailyDialog是一个高质量的多轮开放域英语对话数据集，包含13,118个对话，平均每个对话有约8个发言者轮次和每轮约15个标记，适用于对话文本分类任务。