I am trying to use the new gpt 4 turbo model for web browsing.
I went through their documentation but still i am not clear on what i need to pass as a function in terms of making it access the web.
I am planning to use the SerpAPI for web browsing. So how can i configure the function to do web browsing using SerpAPI using the latest gpt 4 turbo model API.
The “old” way of doing it would be through a multi step process where you would take someone’s reply, extract a plausible set of search terms, then send those searches to SerpAPI to extract contextual information and add it to the prompt (or a vector database to query) for added context…
I believe now via function calling you can accomplish this without building out a code base openai(.)com/docs/assistants/tools/function-calling, but I have yet to set this up myself
Yes thats right. with the latest update in gpt 4 turbo it looks like we can do without any code. But i dont know how.
its same as passing a tool to Langchain agent.
But when i try to define a function its only a json format. I dont understand. How can i integrate this.
