This is a web application built using VanillaJS and Vite, hosted on Vercel. The server is hosted on Render. The application allows the users to ask for help with programming language(s) -- similar to ChatGPT -- such as JavaScript, React, and more.
It has an elegant user interface similar to the ChatGPT app, and it is capable of communicating with the advanced GPT-3 model API (in this example, text-davinci-001 is being used). One of the main features of this application is the ability to ask for help with programming languages such as JavaScript and React. Additionally, it can take code and translate it into another programming language.
A web application built using VanillaJS framework via Vite - version 4.2.0, which is a next generation frontend tooling and hosted using Vercel. The web service (server) was hosted on Render.
The following are the server packages:
| Package Name | Version |
|---|---|
| cors | ^2.8.5 |
| dotenv | ^16.0.3 |
| express | ^4.18.2 |
| nodemon | ^2.0.21 |
| openai | ^3.2.1 |
You can lauch this application ---> here [*my token has reached the limit].
Make sure you clone the repository from the main branch.
git clone git@github.com:sangay-yonten/OpenAI-Codex.git
After cloning the repository, perform the following to start the Frontend application and Backend server.
(a) STEP 1: Start the Frontend (web application)
cd client
npm install && npm run dev
(b) STEP 2: Open split terminal and start the server (BE)
cd server
npm install && npm run server
(c) STEP 3: To add your custom API key and change the model
In the server directory, add a .env file and write OPENAI_API_KEY = "api_value_here", then save it and re-start or re-run the server.
Open server.js file and update the model value/name by the desired model and update the changes to use it.
So to make use of the above changes, you must last update the handleSubmit function in script.js file, which fetches the API response from frontend.
Replace 'https://openai-codex-57kl.onrender.com' with the value port-the-server-listens-to, in my case I have kept it as,
http://localhost:4200, when I start the server locally.
If you're experiencing 401, 429, or similar issues, here are some potential cause and solutions:
- Proper API key from OpenAI: It's possible that you are not using a valid API key from OpenAI. You can get an API key by visiting https://beta.openai.com/account/api-keys.
- Free credits offered by OpenAI: If you are using the free credits offered by OpenAI to use their models, you can check your usage at https://beta.openai.com/account/usage. If you have used up your free credits, you can check the pricing and upgrade your account at https://openai.com/api/pricing/.
- Proper server deployment: Make sure you use the deployed link from Render at https://render.com/. If you are unsure how to deploy the project, follow the steps mentioned in the video.
- After successfully deploying the link, make sure to paste the link in the script file of the client and push it to GitHub. This will allow Vercel to detect and deploy the application properly. Also, please note that the server link used in the Codex repository shared on GitHub has used all of its credits, so it will throw a 429 error and will not work. You will need to use your own server link by deploying it on Render or using another hosting service.
You can even contact us for more details and open discussions.