Could you use namespaces to weigh web search results for providing answers to LLM queries? When you originally asked about the most recent GPTuesdays that you missed, could it have used a namespace you previously generated, to search the web to try to find out about the most recent GPTuesday?
Yes that would work too... An LLM agent can be given the capability to connect to a search engine like Google via an API Like this one for example: serpapi.com/ ChatGPT I think connects to Bing when answering many questions, etc. That could/would work but there still might be info that you don't want on the web or want more control over... That's where a vector db comes in ie: Think a company's internal documents etc.
Thanks
Could you use namespaces to weigh web search results for providing answers to LLM queries? When you originally asked about the most recent GPTuesdays that you missed, could it have used a namespace you previously generated, to search the web to try to find out about the most recent GPTuesday?
Yes that would work too...
An LLM agent can be given the capability to connect to a search engine like Google via an API
Like this one for example: serpapi.com/
ChatGPT I think connects to Bing when answering many questions, etc.
That could/would work but there still might be info that you don't want on the web or want more control over...
That's where a vector db comes in
ie: Think a company's internal documents etc.
You have the shirt, but do you have the socks?
"How can I help?" socks
No do you? : )