selfcorrexp/llama3_additional_rr40k_non_delete_sft_chat_format
收藏Hugging Face2024-12-23 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/selfcorrexp/llama3_additional_rr40k_non_delete_sft_chat_format
下载链接
链接失效反馈官方服务：
资源简介：
这是一个包含对话内容的训练数据集，数据集中的每个样本包括索引、提示、答案、是否为第一轮对话、真实标签、奖励、解决方案、是否达到目标标志位、对话轮次以及具体的对话内容（包括对话文本和对话角色）。数据集仅包含训练集划分，且提供了数据集的字节数和示例数量。
This is a training dataset containing conversation content. Each sample in the dataset includes an index, a prompt, answers, whether it is the first round of conversation, ground truth labels, rewards, solutions, a flag indicating whether the goal is met, the turn of the conversation, and specific conversation content (including text and the role of the speaker). The dataset is split only into a training set and provides the number of bytes and examples.
提供机构：
selfcorrexp
