AI模型精选76°

Mistral发布开源内容安全模型Shieldstral,3B参数可端侧部署

🛡️Introducing Shieldstral, Mistral’s 3B open-weights model for content safety that can be deployed ...

精选理由

Mistral出了个3B的开源内容安全模型,能直接在手机上跑,适合做本地内容审核。

AI 摘要

Mistral发布Shieldstral,这是一款30亿参数的开源权重内容安全模型。该模型专为检测有害内容设计,可部署在手机等终端设备上。Shieldstral权重完全开放,开发者能在本地运行并集成到应用中。端侧部署意味着审核可在设备本地完成,减少对云端的依赖。

原文 · Mistral AI

🛡️Introducing Shieldstral, Mistral’s 3B open-weights model for content safety that can be deployed ...

🛡️Introducing Shieldstral, Mistral’s 3B open-weights model for content safety that can be deployed on-device � mistral.ai/news/shieldstr… 9p 💬 13 🔄 14 ❤️ 76 👀 4330 📊 25 ⚡