Using vector database tables with ChatGPT retrieval plugin

GPT builders Plugins / Actions builders
,
1

Hi fellas, I am using a Postgres database for the ChatGPT retrieval plugin, and I created the documents table, the indices and the match_page_sections function according to migrations sql

I am able to insert a pdf file using:

curl -X POST -F “file=@./postgresql-15-US.pdf” http://localhost:3333/upsert-file

and using http://localhost:3333/docs#/default/query_main_query_post I can query:

{
      "queries":[
        {
            "query":"find when should I use check",
            "top_k":3
        },
        {
            "query":"find when should I use using",
            "top_k":3
        }
        ]
}

All of this works fine using the documents table by default. My questions are:

  1. Should I create additional vector tables to support ChatGPT retrieval plugin?
  2. Having populated other vector tables, how do I redirect the ChatGPT retrieval plugin to a particular table?
  3. How does the code that queries OpenAI take the plugin as a parameter?

Thanks in advance

1 Like