abhishekgahlot/GLM-5.1-1000000x

GLM-5.1-1000000x是一个由GLM-5.1模型生成的推理轨迹数据集，包含1,003,589个条目。每个条目包含完整的思维链（chain-of-thought）推理轨迹和最终答案。数据集分为四个子集：main（通用推理和指令遵循，占59.6%）、Math（数学，占20.8%）、PHD-Science（研究生级物理、化学、生物，占10.3%）和Multilingual-STEM（多语言STEM，占9.3%）。数据集总估计token数约为5.36B，平均每个记录约5,338个token。支持英语和中文，主要用于文本生成和问答任务。

GLM-5.1-1000000x is a dataset of reasoning traces generated by the GLM-5.1 model, containing 1,003,589 entries. Each entry includes a full chain-of-thought reasoning trace followed by the final answer. The dataset is divided into four subsets: main (general reasoning & instruction-following, 59.6%), Math (mathematics, 20.8%), PHD-Science (graduate-level Physics, Chemistry, Biology, 10.3%), and Multilingual-STEM (STEM in Chinese, English & other languages, 9.3%). The total estimated tokens are ~5.36B, with an average of ~5,338 tokens per record. It supports English and Chinese and is primarily used for text-generation and question-answering tasks.