cazonai/elyra-0.3-dataset

Elyra 0.3是一个用于微调巴西人工智能Elyra 0.3的数据集，由Cazon AI开发。数据集结合了5层丰富内容和AutoRadio基础数据，包括Cazon AI的历史代码、巴西日常生活（如Pix、INSS、IR等）、巴西文化（如节日、食物、足球等）、安全保护措施（如心理健康、暴力、LGPD等）和巴西个性（如自然填充、友好对话等）。总样本数约为24-25k，格式为ChatML格式的JSON。数据集旨在支持Elyra的持续学习和进化，强调巴西日常生活的真实性和伦理保护。

Elyra 0.3 is a dataset for fine-tuning the Brazilian AI Elyra 0.3, developed by Cazon AI. The dataset combines 5 layers of enrichment + AutoRadio base, including Cazon AIs historical codices (Elyra/Sirena/AutoRadio + Maestro), Brazilian daily life (Pix, INSS, IR, MEI, Detran, FGTS, etc.), Brazilian culture (festivals, food, football, music, regions), safeguards (mental health, violence, LGPD, resources), and Brazilian personality (natural fillers, short conversations, friendly tone). The total estimated samples are ~24-25k in ChatML format. The dataset supports Elyras continual learning and evolution, emphasizing the authenticity of Brazilian daily life and ethical safeguards.