From the course: Building Secure and Trustworthy LLMs Using NVIDIA Guardrails
Join today to access over 24,000 courses taught by industry experts.
Enhancing LLMs with custom actions
From the course: Building Secure and Trustworthy LLMs Using NVIDIA Guardrails
Enhancing LLMs with custom actions
- [Narrator] Let's take a look at an exciting aspect of large language models, custom actions within guardrails. Our focus will be on understanding what custom actions are, how they work, and why they are a powerful tool for extending the capabilities of large language models. You'll have a clear grasp of how to use these actions to innovate and create more versatile user-centric AI solutions. So let's get started. So what exactly are custom actions In the context of LLMs, custom actions are specific tasks that models can execute by calling external functions. Think of them as additional skills you can equip your LLM with, allowing it to perform specialized tasks beyond its review fault capabilities. For example, you may have an action that calls an external API to fetch the current location or weather or another that performs complex calculations. These actions enable LLMs to go beyond generating text to actually interacting with other systems and data sources, thus enhancing their…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.