Strange Answer form Gala agent

Hi , i’m doing the Unit3 of agent course, about USE CASE FOR AGENTIC RAG. I’m using a different llm with respect the one in the tutorial. Actually, I’m using a local LLM “llm = Ollama(model=“llama3.2”, request_timeout=360.0)”. I can’t understand why it gives me some strange aswers when I ask for informations that should be retrived from a db. For example , for query “Tell me about our guest named ‘Lady Ada Lovelace’.”, it aswers :

"Lady Ada Lovelace was a mathematician and the world’s first computer programmer. She is best known for her work on Charles Babbage’s Analytical Engine, a mechanical general-purpose computer. Lady Ada Lovelace made significant contributions to mathematics and computing, and her notes on the engine are considered to be the first program for a computer.

In addition to her work on the Analytical Engine, Lady Ada Lovelace was also known for her strong support of Charles Babbage’s work and her vision for its potential. She believed that machines could go beyond mere calculation and perform any task that humans can, which was a revolutionary idea at the time.

Unfortunately, I do not have information linking this lady with Marie Curie"

even if Marie Curie isn’t mentioned in the question.

Again, for the query “What is Facebook and what’s their most popular model?”, it answers:
":top_hat: Alfred’s Response:
To answer the original user question:

Facebook is a social media platform that allows users to create profiles, connect with friends and family, share updates and photos, and join groups based on common interests. It was founded in 2004 by Mark Zuckerberg and has since become one of the most widely used social media platforms in the world.

As for their most popular model, I couldn’t find any information on a specific “model” developed by Facebook. However, the company does use various machine learning models to power its services, such as the ESM (Efficient Short-Term Monolingual) transformer, which is used for natural language processing tasks like text classification and question answering.

If you’re looking for more general information on Facebook’s features or history, I’d be happy to help!", even if it should use the necessary tool and return the actual most popular model.

Can anyone explain me the reason for these answers?

1 Like