Skip to content

Latest commit

 

History

14 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 

Repository files navigation

Last updated 02/20/26

UVA Library Workshop on Python Programming with Local LLMs and Agentic AI

Link to recording of this workshop

About Me

  • Erich Purpur: I am a Research Librarian for Science & Engineering in the Brown Science & Engineering Library. I've been at UVA since 2017.
  • epurpur@virginia.edu

Data Resources in the UVA Library

Workshop Calendar

Workshop Date Time
Intro to Python pt 1 Tuesday 1/20 3:00 - 4:30pm
Intro to Python pt 2 Tuesday 1/27 3:00 - 4:30pm
Python & APIs Tuesday 2/3 3:00 - 4:30pm
Using Large Lanugage Models Locally Tuesday 2/17 3:00 - 4:30pm
Python & Local LLMs Tuesday 2/24 3:00 - 4:30pm

Disclaimer

I am not an AI expert! We are all learning and navigating this confusing landscape together.

Install Anaconda (this will take some time)

Background Information

Large Language Models

A Large Language Model (LLM) is an artificial intelligence algorithm that uses deep learning techniques and massive data sets to understand and create human-like content across various media types including text, images, video, etc. These models are trained on extensive datasets of diverse information. Using deep learning techniques, they recognize patterns and relationships within and between different forms of media.

In simpler terms, an LLM is a computer program that has been fed enough examples of human language that it can recognize and interpret human language and other forms of complex data. Many LLMs are trained on data that has been gathered from the internet and the quality of the samples impacts how well the LLM will learn natural language.

There are many LLMs in existence such as OpenAI's high profile GPT (Generative Pre-Trained Transformer). Some are optimized for specific purposes, as we will learn more about today, while others are more general use.

LLMs are composed of large files that contain the parameters and architecture defining the model. These can be several gigabytes in size. Once the model is downloaded, it may need to be configured for a specific task.

Why?

With services available like ChatGPT and Copilot, you might be wondering why would you want to run an LLM directly on your own machine? Well, there are many reasons but basically, using local LLMs gives you a level of control and customizability that you don't have otherwise. On top of that, one major use case is for privacy concerns. If you are working with sensitive data, interacting with it locally eliminates the worry it will be exposed to the internet. Other reasons to use local LLMs include (but are not limited to) free cost, offline use, and environmental reasons.

Ollama

Ollama is an open source project that serves as a powerful and user-friendly platform for running LLMs on your local computer. Ollama simplifies the process of downloading, installing, and interacting with a wide range of LLMs. See available models here. Ollama runs as a service on your local computer which you can interact with just like any other.

Install Ollama LLMs

ollama pull llama3.2

ollama pull phi3:3.8b

ollama pull moondream

Agentic AI

In the past few years, AI has increased programming efficiency and productivity hugely. Nowadays, many people are using AI Agents to write code for them. You might have used AI to help write code, which you copy and pasted into your scripts. In this process, the human is still driving the interaction. AI Agents are basically software systems that can perform tasks autonomously by orchestrating multiple step processes. If given a broad outline (prompt) of instructions and access to the required tools, it has agency to decide how to accomplish that task. Human oversight may or may not be involved.

Vibe Coding

Another software/programming paradigm in use today is vibe coding. Vibe Coding is AI-assisted software development where programmers use natural language prompts to guide the AI to create, debug, and refine code. The focus is on the output and functionality of the code or software (the vibe). Rather than line-by-line manual coding as had been done in the past, the developer oversees the AI-created code and then tweaks it accordingly.

Agentic AI Tools

UVA Copilot - Use "agent builder" feature. No installation required. This is more like retrieval augmented generation (RAG) than an AI Agent

Replit - Another browser-based environment, from Base 44. No installation required!

Firebase Studio - Web-based google product, minimal installation required, build and execute everything in a web browser!

Github Copilot - integrate into existing programming environment (VS Code for example). Installation required.

Antigravity - AI Agent IDE for building software projects. Installation required.

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages