jescy525/archon-sft-v1-mix

archon-sft-v1-mix是一个AETHER家族的监督微调数据集，属于mix组。该数据集用于文本生成和对话任务，包含约10万到100万条样本，支持英语和法语。数据格式为JSONL ChatML消息，每个条目包括消息列表（角色为system、user或assistant，以及内容）、任务类型（如函数调用、代码、推理链等）、来源数据集ID、语言和系统来源。数据集应用了MinHash去重技术，阈值为0.85，以提高数据质量。它通过prepare_sft.py管道生成，日期为2026-05-25，适用于聊天机器人和AI模型训练。

archon-sft-v1-mix is a supervised fine-tuning dataset from the AETHER family, belonging to the mix group. It is designed for text-generation and conversational tasks, containing approximately 100,000 to 1,000,000 samples in English and French. The data format is JSONL ChatML messages, with each entry including a list of messages (with roles such as system, user, or assistant, and content), task type (e.g., function_calling, code, reasoning_cot), source dataset ID, language, and system source. MinHash deduplication is applied with a threshold of 0.85 to enhance data quality. Generated via the prepare_sft.py pipeline on 2026-05-25, it is suitable for chatbot and AI model training.