zake7749/kyara-zh-sample-1M

Kyara（知识产出自适应检索增强）是一个通过知识检索过程提高语言模型能力的实验性项目，尤其针对代表性不足的语言，如繁体中文。本项目是Kyara 2.5 SFT数据集的一部分，目的是扩展用于模型训练的繁体中文有限语料库。数据集包含会话信息，每个会话由from和value两个字段组成，均为字符串类型。

Kyara (Knowledge Yielding Adaptive Retrieval Augmentation) is an experimental project aimed at enhancing language models through knowledge retrieval processes, especially for underrepresented languages like Traditional Chinese. This dataset is a part of the Kyara 2.5 SFT dataset, intended to expand the limited Traditional Chinese corpus used for model training. The dataset includes conversational information, with each conversation consisting of from and value fields, both of which are strings.