Triangle104/QWQ-LONGCOT-500K

这个数据集包含约500,000个使用QwQ-32B-Preview语言模型生成的响应实例，通过结合多个高质量来源的提示来创建多样化的综合训练数据。数据集主要是英文的，是合成的，并可能包含不准确的信息，可能无法很好地泛化到现实世界案例。

This dataset contains approximately 500,000 instances of responses generated using the QwQ-32B-Preview language model, created by combining prompts from multiple high-quality sources to produce diverse and comprehensive training data. The dataset is predominantly in English, synthetic in nature, and may contain inaccuracies that may not generalize well to real-world cases.