The Stack Overflow Podcast From prompt attacks to data leaks, LLMs offer new capabilities and new threats
Dec 20, 2023
The podcast discusses the benefits and downsides of large language models (LLMs) and the need for security. It explores adversarial attacks on LLMs, the value proposition of AWS, experiments with fine-tuning LLMs, and the importance of keeping humans at the center of models and giving credit to knowledge contributors.
Chapters
Transcript
Episode notes
1 2 3 4 5
Introduction
00:00 • 3min
Adversarial Attacks and Security Issues
02:48 • 12min
Value Proposition of AWS and ML Deep Learning Models
14:40 • 5min
Experiments with Fine-tuning LLMs and Source Attribution
19:41 • 2min
Keeping Humans at the Center and Giving Credit to Knowledge Contributors
21:34 • 3min