thinkwee/NOVEReason_full

NOVEReason_full是一个多领域、多任务、通用推理数据集，包含七个经过精心策划的子数据集，涵盖一般推理、创意写作、社交智能和多种语言理解等四个子领域。该数据集已经经过仔细清洗和过滤，以适应使用强化学习训练大型推理模型的需要。它包含大约170万个训练示例，适用于问题回答、翻译、摘要和文本生成等多种任务。

NOVEReason_full is a multi-domain, multi-task, general-purpose reasoning dataset, comprising seven curated sub-datasets across four subfields: general reasoning, creative writing, social intelligence, and multilingual understanding. The dataset has been carefully cleaned and filtered to ensure its suitability for training large reasoning models using reinforcement learning. It contains approximately 1.7 million training examples and is suitable for tasks such as question answering, translation, summarization, and text generation.