Azure99/blossom-v6.2-sft-stage2

BLOSSOM V6.2 SFT Stage2是一个为Blossom V6.2模型第二阶段SFT训练设计的高质量、多样化的语言模型微调数据集。它主要包含中文和英文数据，目的是进一步提高模型处理复杂指令的能力，尤其是在处理罕见现实世界问题时。数据集由ShareGPT、WildChat、Wizard、Stackoverflow、Math等来源合成，通过三种成本效益高的模型进行响应生成，并经过N-Gram过滤和毒性内容过滤。每个条目代表一个会话样本，包含唯一标识符、类型、来源和对话消息。

BLOSSOM V6.2 SFT Stage2 is a high-quality, diverse language model fine-tuning dataset designed for the second-stage SFT training of the Blossom V6.2 model. It primarily contains Chinese and English data, aiming to further enhance the models ability to handle complex instructions, especially in rare real-world scenarios. The dataset is synthesized from sources like ShareGPT, WildChat, Wizard, Stackoverflow, Math, etc., using three cost-effective models for response generation, and is filtered through N-Gram and toxic content filters. Each entry represents a conversational sample with a unique identifier, type, source, and dialogue messages.