tokyotech-llm/swallow-math
收藏Hugging Face2026-03-01 更新2025-05-31 收录
下载链接：
https://hf-mirror.com/datasets/tokyotech-llm/swallow-math
下载链接
链接失效反馈官方服务：
资源简介：
SwallowMath是一个包含大约23亿个标记的高质量数学数据集，由FineMath-4+数据集经过LLM驱动的重写管道生成，使用Llama-3.3-70B-Instruct模型进行重写，去除了模板内容，恢复了缺失的上下文，并将解决方案重写为简洁的逐步解释。该数据集适合训练大型语言模型进行数学任务。
SwallowMath is a high-quality mathematical dataset comprising approximately 2.3 billion tokens, derived from the FineMath-4+ dataset through an LLM-driven rewriting pipeline using the Llama-3.3-70B-Instruct model. It removes boilerplate, restores missing context, and rewrites solutions into concise, step-by-step explanations, making it suitable for training large language models for mathematical tasks.
提供机构：
tokyotech-llm
