ReactiveAI/smol-smoltalk-mini-Interaction-SFT

ReactiveAI/Smol-Smoltalk-Mini Interaction SFT数据集是从HuggingFaceTB/smol-smoltalk派生出的，专为交互式监督微调设计的，适用于Reactive Transformer模型。该数据集只包含单个交互，不使用长期记忆，而是使用短期记忆存储之前的交互信息。数据集包含的每个交互被处理为独立的query和answer对。该数据集适用于小规模研究模型的训练，不适合生产模型使用。

The ReactiveAI/Smol-Smoltalk-Mini Interaction SFT dataset is derived from HuggingFaceTB/smol-smoltalk and is designed for interactive supervised fine-tuning, especially for Reactive Transformer models. This dataset contains only single interactions, does not use long-term memory, and instead uses short-term memory to store information from previous interactions. Each interaction in the dataset is processed as a separate pair of query and answer. The dataset is suitable for training small-scale research models and is not recommended for use in production models.