OctoMed/Opus-Reasoning-Filtered

OctoMed/GLM-5.1-Reasoning是一个用于单轮指令跟随的示例数据集，包含显式的思维链推理，并已转换为OctoMed格式用于SFT训练。数据集来源于Jackrong/GLM-5.1-Reasoning-1M-Cleaned，原始数据的收集、从GLM-5.1的蒸馏和清理工作归功于原始作者。每个示例包含三个字段：question（指令/问题文本）、responses（包括<think>推理块和答案的完整模型输出）和answer（从</think>标签后提取的最终答案）。

OctoMed/GLM-5.1-Reasoning is a dataset of single-turn instruction-following examples with explicit chain-of-thought reasoning, converted to OctoMed format for SFT training. Derived from Jackrong/GLM-5.1-Reasoning-1M-Cleaned, all credit for the original data collection, distillation from GLM-5.1, and cleaning goes to the original authors. Each example contains: question (the instruction / question text), responses (the full model output including <think> reasoning block followed by the answer), and answer (the final answer extracted after the </think> tag).