Adelacici/OpenMementos

OpenMementos-228K是一个包含228,557条推理轨迹的数据集，这些轨迹带有块分割和压缩摘要（mementos）的注释，源自OpenThoughts-v3。该数据集旨在教导语言模型在长形式推理过程中管理自己的上下文。数据集分为数学、科学和代码三个领域，提供了详细的块和摘要大小统计、压缩比率和质量指标。数据集包含两个子集：default用于训练就绪的数据，full包含中间管道输出。README还详细说明了响应格式、特殊标记、数据来源以及用于构建数据集的数据管道。

OpenMementos-228K is a dataset of 228,557 reasoning traces annotated with block segmentation and compressed summaries (mementos), derived from OpenThoughts-v3. The dataset is designed to teach language models to manage their own context during long-form reasoning. It includes examples across math, science, and code domains, with detailed statistics on block and summary sizes, compression ratios, and quality metrics. The dataset is available in two subsets: default for training-ready data and full for intermediate pipeline outputs. The README also explains the response format, special tokens, data sources, and the data pipeline used to construct the dataset.