从发布线程进入 Jev 技术博客
Diogo 在发布线程末尾给出官方技术博客与早期访问入口,适合继续核对模型原理和产品说明。
作者分析 Jev 的开源实现,推荐 featherless-ai/simple-jev 和 zhengxuyu/litjev,可在 Apple Silicon 上本地运行,无需训练,仅需精确提示。
I've analyzed "jev" as open implementations. What comes out of this. Starting from the end. You want to have it running locally today, an intelligent "if": featherless-ai/simple-jev or zhengxuyu/litjev - zero training, precise prompt specification. On Apple Silicon: