kenhktsui/math-classifiers-data
收藏Hugging Face2025-02-26 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/kenhktsui/math-classifiers-data
下载链接
链接失效反馈官方服务：
资源简介：
该数据集用于训练kenhktsui/math-fasttext-classifier模型的预训练数据整理。它包含网络数据和指令响应对的混合，涵盖了来自多个不同来源的数据，包括JeanKaddour/minipile、open-web-math/open-web-math、math-ai/StackMathQA、open-r1/OpenR1-Math-220k、meta-math/MetaMathQA和KbsdJames/Omni-MATH等。数据集中的标签分为Math（数学）和Others（其他），并分为训练集和测试集，其中训练集包含160万条示例，测试集包含40万条示例。
This dataset is used for pretraining data curation for the kenhktsui/math-fasttext-classifier model. It consists of a mix of webdata and instruction response pairs, sourced from various origins such as JeanKaddour/minipile, open-web-math/open-web-math, math-ai/StackMathQA, open-r1/OpenR1-Math-220k, meta-math/MetaMathQA, and KbsdJames/Omni-MATH. The dataset is labeled into Math and Others categories and is split into a training set with 1.6 million examples and a test set with 400,000 examples.
提供机构：
kenhktsui
