Vertex AI Grounding Large Language Models

Grounding allows Google’s large Language models to use your specific data to produce more accurate and relevant responses.

Grounding is particularly useful to reduce hallucinations and answer questions based on specific information the model wasn’t trained on. This approach is also called RAG (Retrieval Augmentation Generation).

Implementing a Grounding architecture can take some time. In fact, I have written a dedicated article on how you can implement your own custom grounding solution.

Generative AI - Document Retrieval and Question Answering with LLMs

Again, thanks to Google, you can now rely on Google Grounding instead of implementing a custom solution (at least for many standard use cases).

Grounding with Vertex AI

Vertex AI Grounding provides two different grounding features.

Ground with Google Search

Allows the Gemini model to use Google search to answer questions. For example, you could ask: “What's the current Google stock price?”.

Ground to your own data

Grounding your own data is based on Vertex AI Search. Your Gemini or PaLM model is accessing Vertex AI Search before processing your prompt and receiving relevant documents.

That means we have two different products involved

1. Vertex AI Gemini or PaLM API
   That provides a large language model which can be used for text or chat.
2. Vertex AI Search
   That provides our grounding data in an efficient way.

Jump Directly to the Code

All the code for this article is ready to use in the GitHub repository. If you have questions, don’t hesitate to contact me via LinkedIn.