Anonymizing Sensitive Data in LLM Prompts - Techniques and Tools

Learn advanced techniques for anonymizing sensitive data in large language model prompts in this 36-minute video tutorial. Explore various anonymization methods and their performance, with a focus on using libraries like Presidio and Outlines. Follow along with hands-on notebook demonstrations that showcase how to implement anonymization using Presidio and the Phi-3 Mini model. Gain insights into working with vLLM, TGI, and GGUF on Mac systems. Access comprehensive resources, including code repositories, slides, and links to relevant tools and models, to enhance your understanding of data privacy in AI applications.

Data Anonymization for LLMs
Anonymisation Performance
Video Overview
Anonymisation Techniques
Libraries - Presidio, Outlines
Notebook Demo: Using Presidio for Anonymisation
Notebook Demo: Using Phi-3 Mini + Outlines
Notes on vLLM, TGI and GGUF on Mac
Resources