DijkstraFTW/ianncity_KIMI-K2.5-1000000x

KIMI-K2.5-1000000x是一个包含1,000,000个推理轨迹的数据集，这些轨迹是从KIMI-K2.5中提取的高质量推理数据。数据集分布广泛，包括编码（50%）、科学（20%）、数学（15%）、计算机科学（5%）、逻辑问题（5%）、创意写作（5%）和多语言STEM（100k完成）。总token数为5B。数据收集使用了修改版的Datagen工具，耗时约80小时。数据集适用于文本生成和问答任务，特别强调推理、思维链、指令调整和SFT。

KIMI-K2.5-1000000x is a dataset containing 1,000,000 reasoning traces distilled from KIMI-K2.5, focusing on high-quality reasoning. The dataset is broadly distributed, including Coding (50%), Science (20%), Math (15%), Computer Science (5%), Logical Questions (5%), Creative Writing (5%), and MultilingualSTEM (100k completions). The total token count is 5B. Data was collected using a modified version of Datagen over approximately 80 hours. The dataset is suitable for text-generation and question-answering tasks, with a particular emphasis on reasoning, chain-of-thought, instruction-tuning, and SFT.