ModelScope· @ModelScope2022 · X·· 2 小时前AI 评分41
AI 导读
ModelScope 上线 Corvus-Gov-3B，一个面向中文政策与公共服务对话的 3B 领域微调模型，基于 Llama 3.2 3B Instruct，用 LLaMA Factory 做 LoRA SFT，基于 100 万条中文政务领域对话样本微调，适配专业术语与应答风格。模型定位政策问答、政务咨询及政企内部助手场景，采用 Apache 2.0 许可。
正文
🏛️ Corvus-Gov-3B brings a compact, domain-tuned model to Chinese policy and public-service dialogue.
🤖 https://modelscope.ai/models/JavaFamily/corvus-gov-3B
💬 Targets scenarios such as policy Q&A, public-service consultation, and internal government or enterprise assistants.
📚 Fine-tuned on one million Chinese government-domain dialogue samples to adapt general language capabilities to specialized terminology and response styles.
🧠 This 3B checkpoint is based on Llama 3.2 3B Instruct and trained with LoRA SFT using LLaMA Factory.
📜 Apache 2.0.
来源：ModelScope · x.com