EricLu/SCP-378K

SCP-378K是一个大规模科学问题-解决方案数据集，包含377,705个示例。它是SCP-116K的改进版本，具有显著增强的解决方案覆盖率：每个问题都与从源材料中提取的匹配解决方案配对。数据集覆盖多个科学学科，包括物理学、化学、生物学、医学和数学，针对本科到博士级别的科学内容。与SCP-116K相比，SCP-378K专注于提供完整的带提取标准答案的问题-解决方案对。由于资源限制，此版本不包括DeepSeek-R1生成的响应或推理轨迹。SCP-378K is a large-scale scientific problem-solution dataset containing 377,705 examples. It is an improved version of SCP-116K with substantially enhanced solution coverage: every problem is paired with a matched solution extracted from the source material. The dataset covers multiple scientific disciplines, including physics, chemistry, biology, medicine, and mathematics, targeting undergraduate to doctoral-level scientific content. Compared with SCP-116K, SCP-378K focuses on providing complete problem-solution pairs with extracted standard answers. Due to resource constraints, this version does not include DeepSeek-R1-generated responses or reasoning traces.