microsoft/PhiCookBook

This is a Phi Family of SLMs book for getting started with Phi Models. Phi a family of open sourced AI models developed by Microsoft. Phi models are the most capable and cost-effective small language models (SLMs) available, outperforming models of the same size and next size up across a variety of language, reasoning, coding, and math benchmarks

What it solves

This repository provides a comprehensive collection of resources, tutorials, and code samples for working with Microsoft's Phi family of small language models (SLMs). It addresses the challenge of building and deploying powerful AI applications—including text, vision, audio, and reasoning tasks—on devices with limited computing power, such as edge devices or local PCs.

How it works

The cookbook acts as a central hub for developers to learn how to integrate Phi models into various environments. It covers the entire lifecycle of an AI project: from setting up your development environment and running inference locally or in the cloud, to fine-tuning models for specific tasks and evaluating their performance using responsible AI principles. It provides hands-on code samples for popular frameworks like ONNX Runtime, .NET, and various cloud-based AI platforms.

Who it’s for

This resource is designed for developers and AI engineers who want to build generative AI applications using efficient, high-performance small language models. It is suitable for those looking to deploy AI on everything from mobile devices and local servers to cloud-based infrastructure.

Highlights

  • Extensive library of hands-on code samples for text, chat, vision, audio, and advanced reasoning.
  • Guides for deploying Phi models across diverse platforms, including iOS, Android, NVIDIA Jetson, and local PCs.
  • Practical tutorials on model quantization and fine-tuning using tools like Microsoft Olive, Apple MLX, and Azure Machine Learning.
  • End-to-end examples for building RAG (Retrieval-Augmented Generation) pipelines and function-calling agents.
  • Support for multiple languages and integration with major AI ecosystems like Hugging Face, Ollama, and NVIDIA NIM.

Related

  • Project
    pipeshub-ai/pipeshub-ai

    PipesHub is an open‑source, self‑hostable platform that connects enterprise data sources (Slack, Google Drive, GitHub, etc.) to AI applications. It builds a permission‑aware knowledge graph, stores vector embeddings, and lets any LLM generate explainable answers with citations. The project includes Docker‑Compose deployment, a no‑code agent builder, MCP support, and SDKs for Python, TypeScript, and Go.

  • Project
    dongrixinyu/JioNLP

    A Python library for Chinese NLP preprocessing and parsing, providing tools for data cleaning, entity extraction, and automatic LLM evaluation.

  • Project
    ilysenko/codex-desktop-linux

    A community-maintained Linux distribution of the official ChatGPT desktop app that adds native package support and a wide range of optional system-integration features.

  • Project
    huggingface/smollm

    SmollM is Hugging Face’s open‑source family of tiny yet powerful models: SmolLM‑3 (a 3 B‑parameter, 128 k‑token, multilingual instruction model) and SmolVLM (a compact vision‑language model). All weights, data mixes, and training configs are public, and the models can be loaded with the standard `transformers` library for CPU or modest‑GPU inference.