Magpie-Align/Magpie-Qwen2.5-Pro-300K-Filtered
收藏Hugging Face2024-10-20 更新2024-12-14 收录
下载链接:
https://hf-mirror.com/datasets/Magpie-Align/Magpie-Qwen2.5-Pro-300K-Filtered
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是通过Qwen/Qwen2.5-72B-Instruct模型使用Magpie方法生成的,包含了300K高质量对话数据。数据集的标签包括输入长度、输出长度、任务类别、输入质量、输入难度、最小邻居距离、安全性和奖励等。过滤设置包括输入质量、指令奖励、去除重复和不完整的指令等。数据集的具体信息包括模型名称、数据集名称、类型和描述。
This dataset is generated by Qwen/Qwen2.5-72B-Instruct using the Magpie method, containing 300K high-quality conversation data. The datasets labels include input length, output length, task category, input quality, input difficulty, minimum neighbor distance, safety, and reward. Filter setups include input quality, instruction reward, removal of repetitive and incomplete instructions, etc. The datasets specific information includes model name, dataset name, type, and description.
提供机构:
Magpie-Align



