arcee-ai/DAM

该数据集包含三个子集，分别用于日语语言指令、数学推理和小学数学问题解决。Ichikara子集专注于日语语言指令，MetaMathQA子集用于数学推理，Orca Math子集则包含小学数学问题。数据集的主要用途是用于大语言模型（LLMs）的指令调优和评估，特别是在日语语言处理和数学推理任务上。数据集的创建目的是支持LLMs在日语语言能力和数学推理能力上的训练和评估。

This dataset contains three subsets: Ichikara, MetaMathQA, and Orca Math, designed for instruction tuning and evaluation of large language models (LLMs). Ichikara focuses on Japanese language instruction, MetaMathQA on mathematical reasoning, and Orca Math on grade-school mathematical problem-solving. The dataset is formatted with the Alpaca instruction template and contains 1,729 samples. The language used in the dataset includes Japanese and English. The dataset is intended for instruction tuning, evaluating LLMs performance in Japanese language and math tasks, and training LLMs to handle both linguistic and mathematical problems. The dataset sources include Ichikara, MetaMathQA, and Orca Math, each with its specific licensing details.