ReactiveAI/TinyStories-mini-Interaction-SFT

这是一个基于roneneldan/TinyStories数据集的交互式监督微调数据集，专为Reactive Transformer架构的第二阶段训练而设计。数据集包含query和answer字段，分别代表用户的问题和模型的回答。数据集分为训练集和验证集，大小适中，适用于小型研究模型。数据集是合成的，不包含现实世界的知识，仅包含关于合成简单故事的问答。

This is an interaction supervised fine-tuning dataset based on the roneneldan/TinyStories dataset, designed for the second training stage of the Reactive Transformer architecture. The dataset contains query and answer fields, representing the users question and the models response, respectively. It is split into training and validation sets, with a moderate size suitable for small research models. The dataset is synthetic and does not contain real-world knowledge, only including Q&A about synthetically generated simple stories.