遇见数据集

Kazki/Magpie-Tanuki-8B-annotated-96k

收藏
Hugging Face2025-02-11 更新2025-02-15 收录
官方服务:

资源简介:

Magpie-Tanuki-8B-annotated-96k数据集是基于Magpie方法对Tanuki-8B-dpo-v1.0数据集进行应用生成的,并使用cyberagent/calm3-22b-chat模型对指令进行难易度、质量、类别注释的数据集。该数据集包含了经过注释的指令,用于评估用户查询的清晰度、具体性和一致性,并为指令分配难易度和类别标签。

The Magpie-Tanuki-8B-annotated-96k dataset is generated by applying the Magpie method to the Tanuki-8B-dpo-v1.0 dataset and annotated for difficulty, quality, and category using the cyberagent/calm3-22b-chat model. This dataset includes annotated instructions for evaluating the clarity, specificity, and consistency of user queries, and assigns difficulty and category labels to the instructions.

提供机构:
Kazki
二维码
社区交流群
二维码
科研交流群
商业服务