What is InstructLab and how do you use it? 9 quick videos to help you get started

InstructLab is a core element of Red Hat Enterprise Linux AI. It's an open source community project that provides a simpler and more accessible way to improve a large language model (LLM) used in generative artificial intelligence (gen AI) applications.

Launched by Red Hat and IBM at the 2024 Red Hat Summit, InstructLab not only brings a truly open source development model to LLMs, it also gives people with minimal technical experience a way to contribute to AI model development and continual improvement.

Join the InstructLab community

But you might be asking yourself: How does it all work? Why is it important? How can you get involved? You're in luck. The InstructLab community has been creating loads of video content to help you get started.

1. What is the InstructLab community project?

In this video, Máirín Duffy, software engineering manager at Red Hat, explains what InstructLab is and how it's revolutionizing AI development by democratizing AI development and training processes. This platform allows anyone, regardless of technical background, to contribute their knowledge and skills to AI models. Here you'll learn how InstructLab's open source workflow helps break down barriers, foster diversity and reduce biases in AI, all while producing community-driven models under open source licenses.

2. The science behind training a model with InstructLab

Here Máirín discusses how InstructLab enhances a large language model (LLM) using far less human-generated data and fewer computing resources by retraining models through the innovative LAB (Large-scale Alignment for chatBots) method. Developed by the MIT-IBM Watson AI Lab and IBM Research, the LAB method enables continuous improvement of AI models through upstream contributions. In this video you'll learn how the LAB method is used for AI model enhancement.

3. Joining the InstructLab community

Wondering how you can contribute to the InstructLab project? In this video, Máirín covers everything you need to get started, including how to contribute to open source AI models without being a machine learning expert. You'll also find out how you can share your skills and knowledge about topics you love, such as bird identification or woodworking, using InstructLab's step-by-step GitHub guide.

4. InstructLab Demo: Lowering the barrier to AI model development

In this video, Jeremy Eder, Distinguished Engineer at Red Hat, goes through an in-depth demo of InstructLab, including how you can correct and enhance AI models without extensive technical knowledge. This video demonstrates the power of InstructLab's synthetic data generation and contributor-friendly workflow, showcasing how even everyday users can improve a model's accuracy and reduce its bias.

5. How to train an LLM using InstructLab

In this demo video, Grant Shipley, a Senior Director at Red Hat, shows you how you can use InstructLab to revolutionize your AI development process. Learn how to serve and interact with LLMs using the open source Granite model, and how to correct model inaccuracies through a simple Q&A format and synthetic data generation. Grant's demo includes training a model with specific knowledge and applying it to real-world tasks.

6. InstructLab: From 0 to aligned model using Mac OS

Here Grant walks you through how to use InstructLab on a Mac. In this demo, Grant covers installing the InstructLab command line tools, and adding knowledge to a model and using a variety of functions. He also goes over how to integrate InstructLab into your workflow, from setup to training, on consumer-grade hardware.

7. Using Windows with InstructLab with Nvidia CUDA support

In this video Grant demonstrates how to use the InstructLab project on Windows through the Windows Subsystem for Linux (WSL). He covers the setup process, including installing Python, creating a virtual environment and configuring InstructLab. If you have an Nvidia GeForce driver, he also shows you how to get InstructLab working with Nvidia CUDA support.

8. InstructLab and Distrobox: A marriage made in heaven

If you use a version of Linux that doesn't include all of the required dependencies for running InstructLab, this video is for you. Grant demonstrates how he runs InstructLab on his Linux gaming machine by using a program called Distrobox to run a Fedora 39 image in a container. He also shows you how to set up NVIDIA GPU acceleration.

9. Using AnythingLLM as a local chat/RAG interface for "ilab model serve" from the InstructLab Project

Not everyone is going to love the built-in InstructLab user interface (UI), so in this video, Grant shows you how to hook up InstructLab to the open source AnythingLLM tool. This tool is available on Mac, Windows and Linux and lets you interact with InstructLab through a more polished UI, and to add retrieval-augmented generation (RAG) capabilities to existing models. Grant goes through how to get and install AnythingLLM on Mac, how to set up and configure a new chat workspace and how he adds RAG capabilities to a model he's trained in the past.

Wrap up

And there you have it, 9 short videos from the InstructLab community to get you started on your AI model training journey. If you have any questions or would like to help out, visit any of the many InstructLab community collaboration spaces.

Learn more about Red Hat Enterprise Linux AI