<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>TFLite on PlumePHP</title><link>https://plumephp.com/tags/tflite/</link><description>Recent content in TFLite on PlumePHP</description><generator>Hugo</generator><language>zh-CN</language><lastBuildDate>Sun, 27 Sep 2026 11:00:00 +0800</lastBuildDate><atom:link href="https://plumephp.com/tags/tflite/index.xml" rel="self" type="application/rss+xml"/><item><title>边缘端推理部署：TFLite、ONNX Mobile 与 NPU 异构加速实践</title><link>https://plumephp.com/ai-edge-inference-deployment/</link><pubDate>Sun, 27 Sep 2026 11:00:00 +0800</pubDate><guid>https://plumephp.com/ai-edge-inference-deployment/</guid><description>&lt;blockquote&gt;
&lt;p&gt;当推理发生在手机、摄像头、车载设备上，约束从&amp;quot;算得快&amp;quot;变成&amp;quot;在有限功耗和内存内算得好&amp;quot;。本文聚焦边缘端推理：对比 TFLite / ONNX Runtime Mobile / ExecuTorch / OpenVINO 等移动框架，讲解 Qualcomm 与 Apple NPU 的异构调度，给出端侧量化与模型压缩的组合拳，并拆解目标检测、端侧 LLM、语音助手三类典型应用。&lt;/p&gt;</description></item></channel></rss>