Custom-ChatGPT with OpenAI API's

This repository contains code for how to store and query your own data using OpenAI Embeddings and Supabase using JavaScript.

Prerequisites

Before you begin, make sure you have the following set up:

• Node.js installed
• Supabase account with a configured database
• OpenAI API key

Installation

1. Clone the repository:

   git clone https://github.com/muzammildafedar/custom-chatgpt-using-js.git
   cd your-repo

2. Install dependencies:

   npm install openai
   npm install @supabase/supabase-js

Configuration

Replace the placeholder values in the code with your actual Supabase and OpenAI credentials:

const supabaseUrl = 'Your Supabase URL';
const supabaseKey = 'Your Supabase API Key';
const openaiConfig = {
    apiKey: 'Your OpenAI API Key',
};

Code Functions

storeEmbedding(title, body, embedding) Description: Stores the provided title, body, and embedding in the Supabase database.

queryEmbeddings(query, matchThreshold, matchCount) Description: Queries Supabase for embeddings that match the provided query, threshold, and count.

getAnswer(query, posts) Description: Generates an answer from OpenAI GPT-3.5 based on the provided query and matched documents.

DB Schema

Create table to store embeddings

create table posts (
  id serial primary key,
  title text not null,
  body text not null,
  embedding vector(1536)
);

In order to "hook up" OpenAI to our embeddings we need to create a function in Postgres find the closest matching values when given a vector.

create or replace function match_posts (
    query_embedding vector(1536),
    match_threshold float,
    match_count int
  )
  returns table (
    id bigint,
    body text,
    title text,
    similarity float
  )
  language sql stable
  as $$
    select
      posts.id,
      posts.body,
      posts.title,
      1 - (posts.embedding <=> query_embedding) as similarity
    from posts
    where 1 - (posts.embedding <=> query_embedding) > match_threshold
    order by similarity desc
    limit match_count;
  $$;

Additionally we can create an index on our posts table to speed up the query.

create index on posts using ivfflat (embedding vector_cosine_ops)
with
  (lists = 100);