TeichAI/DeepSeek-v4-Flash-Chat

DeepSeek v4 Flash Chat数据集是一个用于监督微调的对话数据集，包含6313个训练示例。该数据集由TeichAI使用teich工具生成，所有助手回复均来自deepseek/deepseek-v4-flash模型。数据格式为每行一个JSON对象的换行分隔文件，每个示例包含完整的对话消息序列（包括系统、用户和助手角色），以及辅助字段如系统提示、用户提示、思考过程和回复内容。数据集还包含模型使用情况和元数据信息，适用于训练和微调对话AI模型。

The DeepSeek v4 Flash Chat dataset is a conversational dataset for supervised fine-tuning, containing 6313 training examples. It was generated using the teich tool by TeichAI, with all assistant responses produced by the deepseek/deepseek-v4-flash model. The data is in newline-delimited JSON format where each line is a training example, including complete message sequences (system, user, and assistant roles) along with convenience fields such as system prompt, user prompt, thinking process, and response. The dataset also includes model usage statistics and metadata, making it suitable for training and fine-tuning conversational AI models.