Hi!

Introducing the Phi-3 Small Language Model

Hi!

Phi-3 is an amazing Small Language Model. And hey, it’s also an easy one to use in C#. I already wrote how to use it with ollama, now it’s time to hit the ONNX version.

Introduction to Phi-3 Small Language Model

The Phi-3 Small Language Model (SLM) represents a significant leap forward in the field of artificial intelligence. Developed by Microsoft, the Phi-3 family of models is a collection of the most capable and cost-effective SLMs available today. These models have been meticulously crafted to outperform other models of similar or even larger sizes across various benchmarks, including language understanding, reasoning, coding, and mathematical tasks.

Phi-3 models are not only remarkable for their performance but also for their efficiency and adaptability. They are designed to operate across a wide range of hardware, from traditional computing devices to edge devices like mobile phones and IoT devices. This makes Phi-3 particularly suitable for developers looking to integrate advanced AI capabilities into applications that require strong reasoning, limited compute resources, and low latency.

C# Phi3-Labs GitHub Repository

The GitHub repository in question serves as a practical guide for developers looking to harness the power of the Phi-3 SLM within their applications. It provides a set of sample projects to demostrante how to use Phi-3 and C#.

• LabsPhi301. This is a sample project that uses a local phi3 model to ask a question. The project load a local ONNX Phi-3 model using the Microsoft.ML.OnnxRuntime libraries.
• LabsPhi302. This is a sample project that implement a Console chat using Semantic Kernel.
• LabsPhi303 (Comming soon!). This is a sample project that uses Phi-3 Vision to analyze images.

The repository contains detailed instructions for setting up the development environment, cloning the necessary models from Hugging Face, and installing the Phi-3 model. 

You can learn more about Phi-3 in Hugging Face.

A final running application will look like this.