订阅内容

It feels like 2024 was the Year of artificial intelligence (AI), quickly going from being an interesting experiment to seemingly the only thing anyone was talking about. And it can be hard to keep up with all the news and advancements being made, but hopefully this will help. In these 11 short videos, Red Hatters cover a variety of topics from open source AI, to the new InstructLab project, through identifying which large language model (LLM) is right for your organization, and more.

Try Red Hat Enterprise Linux AI

Grab a cup of coffee and catch up on some of what Red Hat has been up to in the world of AI.

Open source AI is community built AI

Frank La Vigne explains why using fine-tuning technologies like InstructLab and open source models like Granite makes it easier for developers and people with other specialized knowledge to contribute to generative AI (gen AI) models.

InstructLab 101

Not really sure what InstructLab is? In this 100 second intro video Legare Kerrison walks through what it is, how it works, and how you can get started today.

Which LLM is right for you?

There are a bewildering number of LLMs out there already, and it seems like more are being released every day. So how do you begin to evaluate which LLM is best for you, your projects and your business?

Taylor Smith discusses some of the more important factors to keep in mind when looking at LLMs, including whether it has transparent data sources and the ability to scale and perform well in a small form factor.

With LLMs, does size matter?

While LLMs might be what everyone's been talking about lately, we're already starting to bump up against some challenges with them, including their resource-intensive training, scalability issues and data limitations that are threatening to stall future development. But do LLMs all have to be…large?

Cedric Clyburn talks about how larger isn't always better, and how small, specialized models can be more effective and efficient over all.

RAG vs. fine tuning: Different tools for different jobs

Retrieval augmented generation (RAG) is a popular way to inject domain-specific data into LLMs, but it's not a universal solution. Fine-tuning your models can be more effective in certain scenarios. Cedric explains the difference between these two approaches, and talks about where RAG and fine-tuning work best.

Demo: Red Hat documentation chatbot using RAG

Guillaume Moutier demonstrates how Red Hat customers can use gen AI, LLMs and RAG to extract valuable data out of existing documentation, knowledge bases and other data sources to provide accurate, comprehensive and context-aware answers.

The full model workflow makes use of an array of Red Hat products including:

These are used to augment an LLM with documentation data that's pulled from an open source vector database and served on a single stack model server, all within OpenShift AI.

Red Hat OpenShift AI: Features and architecture

Christopher Nuland provides a technical overview of OpenShift AI, which extends the capabilities of Red Hat OpenShift to develop, train, serve and manage AI models. He highlights the end-to-end machine language operations (MLOps) features and tooling for both predictive AI and gen AI that run on-premise or in cloud environments. He also goes over more advanced features like distributed workloads, GPU accelerator support and data science pipelines.

Red Hat OpenShift AI: Predictive and generative AI demo

Expanding on his previous video, Christopher builds an OpenShift AI demo that combines gen AI (for summarization and sentiment analysis) with predictive AI (for vehicle image detection) to create an insurance claims processing application. He walks through how to use OpenShift AI for model training and experimentation, including how to create data science projects, import data, load Jupyter notebooks and train the models.

AI inferencing at the edge using OpenShift AI

Deploying AI at the edge can help provide organizations with real-time insights into their data. OpenShift AI can be used to deploy models, predict failures, detect anomalies and do quality inspection in low-latency environments in near real-time.

Myriam Fentanes and Landon LaSmith demonstrate how an AI model can be packaged into an inference container image and uses data science pipelines to fetch models, as well as build, test, deploy and update them using a GitOps workflow.

AI-assisted farming

Through a precision agriculture example, Guillaume demonstrates how AI can be used to help farmers increase crop yield and lower costs. Using a trained computer vision model, this game-like demonstration simulates drones flying over crop fields to detect whether they are diseased or healthy. If a diseased field is found, tractors are directed on the shortest path to that field to begin treatment. This demo combines telco 5G slices, edge computing, computer vision and operations research, all orchestrated on OpenShift and developed with the help of OpenShift AI.

Watch more

Want to see more? We have a bunch of YouTube playlists for you to explore.


关于作者

Deb Richardson joined Red Hat in 2021 and is a Senior Content Strategist, primarily working on the Red Hat Blog.

Read full bio
UI_Icon-Red_Hat-Close-A-Black-RGB

按频道浏览

automation icon

自动化

有关技术、团队和环境 IT 自动化的最新信息

AI icon

人工智能

平台更新使客户可以在任何地方运行人工智能工作负载

open hybrid cloud icon

开放混合云

了解我们如何利用混合云构建更灵活的未来

security icon

安全防护

有关我们如何跨环境和技术减少风险的最新信息

edge icon

边缘计算

简化边缘运维的平台更新

Infrastructure icon

基础架构

全球领先企业 Linux 平台的最新动态

application development icon

应用领域

我们针对最严峻的应用挑战的解决方案

Original series icon

原创节目

关于企业技术领域的创客和领导者们有趣的故事