Semantic Kernel Example

Welcome to the Semantic Kernel Example repository! Here, you'll find a lightweight Python demo designed to facilitate the learning of the core concept that underlies Semantic Kernel. Before we delve into the code, let's start with a brief introduction to Semantic Kernel.

What is Semantic Kernel?

Semantic Kernel is an orchestration framework in the form of a lightweight SDK that enables developers to combine AI services, such as Azure OpenAI with familiar programming languages such as C# and Python. It serves as an AI orchestration engine to orchestrate the use of AI or conventional plugins/skills, providing a flexible and extensible framework for building AI applications.

Information: The content and media in this README are adapted from the Semantic Kernel documentation. Please refer to the documentation for more details. For more implementation examples, visit Semantic Kernel GitHub repo

Semantic Kernel Building Blocks

Here we discuss three key components of Semantic Kernel.

Plugins/Skills 🧩

Plugins/skills form the foundational building blocks of Semantic Kernel. These modules encapsulate specific functionalities that can be integrated into an application. While LLM-based semantic functions are commonly used, Semantic Kernel also supports native functions implemented in programming languages like C# and Python.

Just providing functions, however, is not enough to make a plugin. To power automatic orchestration with a planner, plugins also need to provide details that semantically describe how they behave.

Planner 📝

Functions provided by the plugins can be chained together in a predefined order to fulfill user requests. However, anticipating all possible user requests can be challenging. This is where the Planner component becomes useful. Planner is a function that takes a user's ask and returns back a plan on how to accomplish the request. It does so by using AI to mix-and-match the plugins registered in the kernel so that it can recombine them into a series of steps that complete a goal. Semantic Kernel provides some out-of-the-box planners, but developers can also create custom planner plugins.

Connectors 🤝

Connectors act as the bridge between Semantic Kernel and external services (APIs) or data sources. These connectors facilitate API interaction or data exchange.

Implementation Example

This repository provides a simple console app chat.py, built using Semantic Kernel. The app can provide information about a fictional company's healthcare plans, employee benefits and workplace safety tips. The following shows an example interaction with the app:

Information: The app is inspired by this OpenAI demo. Check out their code repo to learn more about the implementation pattern of Retrieval Augmented Generation (RAG).

How to Run the App

Follow these steps to run the console application:

1. Clone the Repository: Open your terminal and execute the following command to clone the repository:

   git clone https://github.com/yuj18/semantic-kernel-example.git
   cd semantic-kernel-example

2. Create a Conda Environment: Create a conda environment using the conda.yml file provided in the repository:

   conda env create -f environment/conda.yml
   conda activate sk-example

3. Create an .env File: Create a .env file in the root directory based on the .env.example file provided in the repository.

4. Prepare the Data: Create text chunks using the example PDFs provided in the data/input directory, then proceed to index them using Azure Cognitive Search. In addition to the text chunks, the demo code also embeds text chunks using Azure OpenAI embedding model (text-embedding-ada-002) to enable vector search with Azure Cognitive Search.

   To prepare the data, run the following command:

   cd scripts
   python data_prep.py --data_input_dir ../data/input --data_output_dir ../data/output --category "handbook"

5. Run the App: Run the app using the following command:

   cd src
   python chat.py

Repository Structure

├── data # Data files
│   ├── input # Example pdfs
├── environment 
│   └── conda.yml # Conda environment file
├── .env.example # Example config file
├── scripts 
│   └── data_prep.py # Data preparation script
└── src
    ├── plugins # Custom plugins
    │   ├── knowledge_base_search
    │   │   ├── create_answer
    │   │   ├── create_search_query
    │   │   └── safety_share
    │   ├── planning
    │   │   └── planner
    │   ├── cognitive_search.py
    │   └── orchestrator.py
    ├── utils.py # Utility functions
    └── chat.py # Console app to interact with user

Please feel free to explore the code in the repository and leverage it as a reference for your Copilot development. Happy coding! 🚀