opencompass/CodeForce_SAGA

CodeForce-SAGA是一个高质量的大型代码生成和问题解决训练数据集，来源于Codeforces竞赛平台。它采用SAGA框架，通过人类与LLM的合作生成严格的测试用例，确保每个解决方案都经过功能稳健性验证。数据集包含问题描述、来源、资源限制和测试用例等详细信息，适合用于大型语言模型的监督微调和指令微调。

CodeForce-SAGA is a high-quality, large-scale code generation and problem-solving training dataset sourced from the Codeforces competitive programming platform. It utilizes the SAGA framework to generate rigorous test cases through human-LLM collaboration, ensuring that each solution is functionally robust. The dataset includes detailed information such as problem descriptions, sources, resource limits, and test cases, making it suitable for supervised fine-tuning and instruction tuning of large language models.