jablonkagroup/chempile-code

ChemPile-Code是一个包含来自化学、生物学和材料科学领域的精选科学代码的数据集。它由两个主要子集组成：CodeParrot和StarCoder，这些子集分别从GitHub和Stack数据集中提取并经过关键词过滤。数据集旨在为训练语言模型提供丰富的科学编程资源，并支持代码生成、科学计算、代码搜索、文档生成和领域适应等应用。

ChemPile-Code is a dataset containing curated scientific code from fields such as chemistry, biology, and materials science. It consists of two main subsets: CodeParrot and StarCoder, extracted from GitHub and Stack datasets respectively, and filtered using keyword-based methods. The dataset aims to provide a rich resource for training language models and supports applications in code generation, scientific computing, code search, documentation, and domain adaptation.