Rivet: How To Use Function Calling And Extend ChatGPTs Abilities! - No Code Tutorial

This video shows how to use function calling to enhance ChatGPT with capabilities it would usually not have (e.g. search the web).
Links:
- Get the project file: github.com/ai-made-approachab...
- See my video on semantic routing: • Rivet: Empower Your Ch...
- Get Rivet: rivet.ironcladapp.com/
- Follow me on LinkedIn: / tim-k%c3%b6hler-ai-mad...
- If you enjoy my content, consider buying me a coffee: www.buymeacoffee.com/ai_made_...

[@AIMadeApproachable]

I forgot to explain why this can only be used with ChatGPT:
- Other LLMs are not trained on function calling (especially not on the openAI format)
- To my knowledge there is no LLM that can be run locally that is able to be a normal conversational agent + execute function calling at the same time. E.g. there is gorillaLLM that can ONLY do function calling. So it cannot even say "hello" to you.
- Only candidate might be Google Gemini. I did not try that yet.

[@gigglesmclovin]

There is a technique that gets around this, basically doing 2 llm calls. the first one has a prompt saying something like "here are some function definitions. when presented with this query (add original query), are any of the function calls going to help answer this. you should respong in the following format ......". then evaluate the function call and pipe the response of this with the original query to a new llm query. voila! function calling with any model. I tested this with germini and it works great.