Welcome To The TinyGen Repo!

What is TinyGen?

https://www.loom.com/share/fea73acba34948c09b7e7960dc1d1cd1

Calling TinyGen Through The Production API

https://www.loom.com/share/81cad00f51b04b1ab823cb3f58fda866

Edit: Streaming is now implemented! Use it through the Streamlit UI or find the endpoints on the API documentation page

Useful Links

• TinyGen UI
• Production API Base Url
• Public Postman Collection To Interact With API
• TinyGen Project Overview

Stack

• Python
• FastAPI
• Google Cloud Run (API Deployment)
• Streamlit & Streamlit Cloud (Frontend + Deployment)
• Supabase (LLM Call Storage)

Development

1. Clone the project project to your computer

git clone https://github.com/tawsifkamal/TinyGen.git

2. Navigate to this project in your terminal

cd TinyGen

3. Obtain the secrets. Use your own or contact me!

If using your own, create a .env file and add the following api keys:

GITHUB_API_TOKEN=
SUPABASE_URL=
SUPABASE_KEY=
OPENAI_API_KEY=

• Github API is required to process all of the files from a given repository
• Supabase is used to store all of the prompts, repoUrls, and responses provided to TinyGen
• OpenAI is the provider for all models used for TinyGen (GPT-4 and GPT-3.5)

4. Install Virtualenv if not it does not exist on your local machine:

pip install virtualenv

5. Create a virtual environment

virtualenv venv

6. Activate the virtual environment

source venv/bin/activate

7. Install all necessary packages

pip install -r requirements.txt

8. Start the Fast API Server

uvicorn api.api:app --reload

9. Open another terminal and start the streamlit frontend

streamlit run frontend/🤖_TinyGen.py

I know it's a bit wierd to have a funky emoji for a file name. But that's how Streamlit chose to do its user design.🤷🏽‍♂️ Please let me know if you do have any problems with running the file!

Making API Calls Directly

The following endpoints are available when making api calls. A public workspace on postman is linked here

Full link

https://elements.getpostman.com/redirect?entityId=17857495-f27f8dae-edd0-4a36-8382-809507a65587&entityType=collection

Otherwise, feel free to call the APIs yourself using your preferred API client. The endpoints are listed below:

If using local API, base_url = http://localhost:8000

If using prod API, base_url = https://tiny-gen-zbkbev3qaq-uc.a.run.app

Tip! The docs for the API is located at the root endpoint /

Call TinyGen

Call TinyGen 1.0

POST /tiny_gen_one/call

• Body Schema = {repoUrl: str, prompt: str}

Call TinyGen 2.0

POST /tiny_gen_two/call

• Body Schema = {repoUrl: str, prompt: str}

Fetch Supabase Data

Fetch Past TinyGen 1.0 Calls From Supabase

GET /tiny_gen_one/get_past_calls

Fetch Past TinyGen 2.0 Calls From Supabase

GET /tiny_gen_two/get_past_calls

Docs

Get API Docs

GET /tiny_gen_two/get_past_calls