https://www.hansking.cn/blog A posts created with Next.js and Tailwind.css en-us admin@hansking.cn (HANS) admin@hansking.cn (HANS) Wed, 22 Apr 2026 10:00:00 GMT https://www.hansking.cn/blog/2026/On-deviceAI-duan-ce-tui-li-de-xing-neng-yu-yin-si-quan-heng https://www.hansking.cn/blog/2026/On-deviceAI-duan-ce-tui-li-de-xing-neng-yu-yin-si-quan-heng [{"value":"On-device AI：端侧推理的性能与隐私权衡","url":"#on-device-ai端侧推理的性能与隐私权衡","depth":1},{"value":"一、先决定哪些任务值得放到设备端","url":"#一先决定哪些任务值得放到设备端","depth":2},{"value":"二、先做内存预算，再谈模型大小","url":"#二先做内存预算再谈模型大小","depth":2},{"value":"三、量化不是免费午餐","url":"#三量化不是免费午餐","depth":2},{"value":"四、推理优化的第一目标是减少等待","url":"#四推理优化的第一目标是减少等待","depth":2},{"value":"五、端侧隐私强，但不是天然安全","url":"#五端侧隐私强但不是天然安全","depth":2},{"value":"六、模型文件也需要保护","url":"#六模型文件也需要保护","depth":2},{"value":"七、端云路由需要可解释","url":"#七端云路由需要可解释","depth":2},{"value":"八、我建议用一张设备矩阵收尾","url":"#八我建议用一张设备矩阵收尾","depth":2},{"value":"总结：端侧 AI 的价值来自克制","url":"#总结端侧-ai-的价值来自克制","depth":2}] Wed, 22 Apr 2026 10:00:00 GMT admin@hansking.cn (HANS) AILLMOn-device AI端侧推理性能优化