altaidevorg/thinking-dataset-en

该数据集是一个高质量的数据集合，专门用于训练语言模型在逻辑推理和问题解决任务中的表现。数据集从多个开源数据集中聚合了与推理相关的内容，并进行了去重和格式化处理，以确保数据的多样性和质量。所有数据条目都被重新格式化为统一的结构，并且标签被标准化为单一的<thinking>标签，以保持一致性。这使得该数据集成为微调语言模型在结构化推理和适应性问题解决中的理想资源。

This Dataset is a curated collection of high-quality data designed for training language models in logical reasoning and problem-solving tasks. It aggregates reasoning-related content from multiple open-source datasets, including logical reasoning, critical thinking, and problem-solving tasks. The dataset has been carefully deduplicated to remove redundant or overly similar examples, ensuring diversity and quality. All data entries were reformatted into a unified structure for ease of use, and labels were standardized by replacing reasoning-related tags with a single, broader <thinking> label to maintain consistency throughout. This comprehensive preprocessing makes the dataset an ideal resource for fine-tuning language models in structured reasoning and adaptive problem-solving.