Session 4 : Harnessing Local LLMs: A Practical Guide for Secure Data Science

Presenter Information/ Coauthors Information

Daniel Burkhalter, OrditusFollow

Presentation Type

Oral

Student

No

Track

Tools

Abstract

The adoption of Large Language Models (LLMs) for data analysis often raises privacy and security concerns due to reliance on cloud services. This talk explores running LLMs locally to address these challenges, comparing performance across three different hardware configurations. Attendees will learn about the development of a custom data analysis tool incorporating local LLMs and see a live demonstration of its capabilities. The session emphasizes practical strategies for integrating open source LLMs locally, enabling secure and efficient analysis of sensitive data.

Start Date

2-7-2025 11:00 AM

End Date

2-7-2025 12:00 PM

This document is currently not available here.

Share

COinS
 
Feb 7th, 11:00 AM Feb 7th, 12:00 PM

Session 4 : Harnessing Local LLMs: A Practical Guide for Secure Data Science

Jacks' Place (Room 050)

The adoption of Large Language Models (LLMs) for data analysis often raises privacy and security concerns due to reliance on cloud services. This talk explores running LLMs locally to address these challenges, comparing performance across three different hardware configurations. Attendees will learn about the development of a custom data analysis tool incorporating local LLMs and see a live demonstration of its capabilities. The session emphasizes practical strategies for integrating open source LLMs locally, enabling secure and efficient analysis of sensitive data.