OsakanaTeishoku/test_synthetic_logical_data_magpie_nemotron_1k
收藏Hugging Face2024-07-16 更新2024-07-22 收录
下载链接：
https://hf-mirror.com/datasets/OsakanaTeishoku/test_synthetic_logical_data_magpie_nemotron_1k
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是使用nvidia/Nemotron-4-340B-Instruct模型合成的单轮指令数据集，类似于Aratako的Synthetic-JP-Conversations-Magpie-Nemotron-4-10k数据集。它可以直接用作指令数据，或作为创建DPO数据集的材料。数据集包含id和messages等特征，其中messages包含role和content。数据集主要用于练习，使用的提示语需要逻辑思维能力来解答问题，且提示语等还有改进的空间。
This dataset is a single-turn instruction dataset synthesized using Nvidias Nemotron-4-340B-Instruct model for practice. It includes two main features: id (integer type) and messages (list type, containing role and content, both string types). The dataset is divided into a training set with 1000 samples. The creation method of this dataset is similar to Aratakos Synthetic-JP-Conversations-Magpie-Nemotron-4-10k dataset, and it can be used directly as instruction data or as material for creating DPO datasets.
提供机构：
OsakanaTeishoku
