观看量 625
本视频由 保利威 提供技术支持

[1]在边缘端部署 Phi-4

0311-banner-16X9.jpg

用 Phi-4 玩转 Edge AI 1  - 在边缘端部署 Phi-4

========================================

Playing with Edge AI using Phi-4 1 - Deploying Phi-4 on the Edge

活动简介


作为全新的 AI 模型架构,Phi-4 系列模型最新推出的 Phi-4-mini 和 Phi-4-multimodal 版本,专门优化了在边缘端的应用构建。随着边缘计算的发展,AI 推理已经不再局限于数据中心,更多应用逐步迁移至设备端,这要求模型不仅要轻量化,更要能够在多样的硬件平台上高效运行。Phi-4 模型凭借其高效的推理能力,尤其适用于在有限资源的边缘设备上部署智能应用。

为了帮助开发者掌握在边缘端利用 Phi-4 系列模型构建应用的技术,三月,微软高级云技术布道师 卢建晖 特别带来「用 Phi-4 玩转 Edge AI」系列直播,从如何在边缘端部署 Phi-4,如何利用全新的 Function Calling 为边缘端创建智能体,到如何在边缘端搭建应用,帮助大家更好地理解如何在边缘端应用 Phi-4 模型,利用先进的技术架构,在不同硬件上实现智能化部署与推理。

3月11日,「用 Phi-4 玩转 Edge AI」系列直播第一期,我们将一起学习如何部署 Phi-4-mini 和 Phi-4-multimodal 模型,并通过量化和转换在不同的硬件平台上搭建推理模型。

As a brand-new AI model architecture, the Phi-4 series models, including the latest Phi-4-mini and Phi-4-multimodal, are specifically optimized for building applications at the edge. With the development of edge computing, AI inference is no longer limited to data centers, more and more applications are gradually migrating to edge devices. This requires models to be not only lightweight but also capable of running efficiently on diverse hardware platforms. Phi-4 models, with their efficient inference capabilities, are particularly well-suited for deploying intelligent applications on resource-constrained edge devices.

To help developers master the technology of building applications with Phi-4 series models at the edge, in March, Kinfey Lo, Microsoft Senior Cloud Advocate, will present the "Playing with Edge AI using Phi-4" live-stream series. The series will cover themes from how to deploy Phi-4 on the edge, how to leverage the new Function Calling to create intelligent agents for the edge, to how to build applications at the edge. This will help everyone better understand how to apply Phi-4 models at the edge and achieve intelligent deployment and inference across different hardware using advanced technological architectures.

On March 11th, in the first session of the "Playing with Edge AI using Phi-4" live-stream series, we will learn how to deploy the Phi-4-mini and Phi-4-multimodal models, and build inference models on different hardware platforms through quantization and conversion.

活动时间:2025年3月11日 19:30 - 20:30

活动形式:微软 Reactor 线上直播


课程资料

更多资讯欢迎扫码关注公众号:MSReactor