Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/pwntus/replicate-support-gpt
Get support, in seconds.
https://github.com/pwntus/replicate-support-gpt
Last synced: 2 months ago
JSON representation
Get support, in seconds.
- Host: GitHub
- URL: https://github.com/pwntus/replicate-support-gpt
- Owner: Pwntus
- Created: 2023-03-15T14:36:27.000Z (almost 2 years ago)
- Default Branch: main
- Last Pushed: 2024-01-22T14:46:11.000Z (11 months ago)
- Last Synced: 2024-10-04T16:13:35.077Z (3 months ago)
- Language: TypeScript
- Homepage: https://replicate-support-gpt.vercel.app
- Size: 696 KB
- Stars: 13
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
> Discontinued, not hosted anymore.
# [Replicate SupportGPT](https://replicate-support-gpt.vercel.app/)
Make ChatGPT answer questions based on **your** documentation.
[](https://replicate-support-gpt.vercel.app/)
## How it works
The goal is to make ChatGPT answer questions within a limited context, where the context is a relevant section of a larger documentation. To do this we use [embeddings](https://platform.openai.com/docs/guides/embeddings). In short, embeddings are tokens converted into vectors that can be used to calculate how closely related two strings are. If we split the documentation into chunks and encode them as embeddings in a vector database, we can query relevant documentation chunks later if we use the same encoding on questions. The relevant documentation chunks will then be used as context for a ChatGPT session.
This app is powered by:
⚡️ [Supabase](https://supabase.com/), for vector based database.
▲ [Vercel](https://vercel.com/), a platform for running web apps.
⚡️ Nuxt.js [server-side API handlers](server/api), for talking to the Supabase database.
📦 [Vuetify](https://vuetifyjs.com/en/), a Vue.js component framework for the browser UI.
## Setup Supabase
Supabase supports vectors in their PostgreSQL database. Create an account and execute the following queries:
Enable the vector extension:
```sql
create extension vector;
```
Create a table for the documentation chunks:
```sql
create table documents (
id bigserial primary key,
content text,
url text,
embedding vector (1536)
);
```
Create a PostgreSQL function that uses the `<=>` cosine distance operator to get similar documentation chunks.
```sql
create or replace function match_documents (
query_embedding vector(1536),
similarity_threshold float,
match_count int
)
returns table (
id bigint,
content text,
url text,
similarity float
)
language plpgsql
as $$
begin
return query
select
documents.id,
documents.content,
documents.url,
1 - (documents.embedding <=> query_embedding) as similarity
from documents
where 1 - (documents.embedding <=> query_embedding) > similarity_threshold
order by similarity desc
limit match_count;
end;
$$;
```
## Run it locally
You need an [OpenAI API key](https://platform.openai.com/account/api-keys), a Supabase URL and Supabase API key (you can find these in the Supabase web portal under Project → API). Copy the contents of [.example.env](.example.env) into a new file in the root of your directory called `.env` and insert the API keys there, like this:
```bash
NUXT_OPENAI_API_KEY=
NUXT_SUPABASE_URL=
NUXT_SUPABASE_KEY=
```
Then, install the dependencies and run the local development server:
```bash
npm install
npm run dev
```
Open http://localhost:3000 in your web browser. Done!
## Populate vector database with embeddings
Modify the [scripts/create-embeddings.js](./scripts/create-embeddings.js) script to include URLs of the documentation to create embeddings of:
```js
// Add documentation URLs to be fetched here
const urls = [
'https://replicate.com/home',
'https://replicate.com/docs',
'https://replicate.com/docs/get-started/python'
// ...
]
```
And run it:
```bash
node scripts/create-embeddings.js
```
## One-click deploy
Deploy this project using [Vercel](https://vercel.com?utm_source=github&utm_medium=readme&utm_campaign=replicate-support-gpt):
[](https://vercel.com/new/clone?repository-url=https://github.com/Pwntus/replicate-support-gpt&env=NUXT_OPENAI_API_KEY&env=NUXT_SUPABASE_URL&env=NUXT_SUPABASE_KEY&project-name=replicate-support-gpt&repo-name=replicate-support-gpt)