qwselfcorr/numia_n8_turn1_corr_gen_regular_processed
收藏Hugging Face2025-02-01 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/qwselfcorr/numia_n8_turn1_corr_gen_regular_processed
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含对话信息，每个对话样本包括索引（idx）、地面真实标签（gt）、是否删除第一个回复（first_rm）、第二个回复的奖励（second_rewards）、标志位（flag）、回合数（turn）以及对话内容（conversations）。对话内容进一步包含对话文本（content）和角色（role）。数据集分为训练集，共有129,667个示例，文件大小为217,981,094字节。
The dataset contains conversational information, with each conversation sample including an index (idx), ground truth label (gt), whether to remove the first response (first_rm), reward for the second response (second_rewards), a flag (flag), the number of turns (turn), and the content of the conversation (conversations). The conversation content further includes the text of the conversation (content) and the role (role). The dataset is split into a training set, with a total of 129,667 examples and a file size of 217,981,094 bytes.
提供机构：
qwselfcorr
