gpinterface (Generative Prompt Interface)

What is gpinterface?

With gpinterface, you can easily create API for your prompts.

Demo

A live demo is available at gpinterface.com.

Here are some specific examples of what gpinterface can do:

• Text Summarization
• Translation
• Text Correction

How does it work?

You can select the LLM model and add context. Once deployed, you'll get endpoints for:

1. one-time chat interactions

2. chat session creation

3. ongoing conversations within a session

4. conversation retrieval from a session

Features

It is a powerful tool designed to streamline the testing and deployment of generative prompts across multiple large language models (LLMs). With an easy-to-use web interface, gpinterface enables rapid configuration and experimentation.

• Simple Deployment: Deploy your configured prompts, creating accessible endpoints instantly.
• Easy Prompt Testing: Utilize a web interface to test prompts with various configurations effortlessly.
• Support for Multiple LLMs: Seamlessly integrate and use different Large Language Models to ensure your application is versatile and robust.

Supported LLMs

gpinterface currently supports a variety of leading Large Language Models, including:

• AI21 Jamba
• Anthropic Claude
• Cohere Command
• Google Gemini
• Meta Llama
• Mistral AI Mistral
• OpenAI ChatGPT

This diverse support allows you to choose the best model for your specific needs and requirements.

Don't see your favorite model? Feel free to open PR or contact me!

Develop

Database

The application requires a PostgreSQL database. Start the database using Docker:

cd backend
docker-compose up -d

The backend utilizes Prisma to manage the database schema and migrations. Execute the following commands at database initialization:

npm run prisma:migrate

To seed the database with initial data for each supported Large Language Model, run the following commands:

npx ts-node prisma/seed

Environment Variables

Backend

The backend requires the following environment variables to be set:

CLIENT_URL="http://localhost:3003"
DATABASE_URL="postgresql://postgres:[email protected]:5432/postgres" # can be replaced with your DB endpoint
AI21_API_KEY="YOUR_AI21_API_KEY"
ANTHROPIC_API_KEY="YOUR_ANTHROPIC_API_KEY"
COHERE_API_KEY="YOUR_COHERE_API_KEY"
GOOGLE_API_KEY="YOUR_GOOGLE_API_KEY"
MISTRAL_API_KEY="YOUR_MISTRAL_API_KEY"
OPENAI_API_KEY="YOUR_OPENAI_API_KEY"
AWS_ACCESS_KEY_ID="AWS_ACCESS_KEY" # you need Llama model access in AWS Bedrock
AWS_SECRET_ACCESS_KEY="AWS_SECRET_KEY" # you need Llama model access in AWS Bedrock
JWT_SECRET="SECURE_RANDOM_STRING"
COOKIE_SECRET="SECURE_RANDOM_STRING"
NODE_ENV="development" # for development logging

Ensure these variables are set in a .env file in the backend directory before starting the application.

frontend

The frontend application requires the following environment variables:

NEXT_PUBLIC_API_ENDPOINT="http://localhost:3000"
NEXT_PUBLIC_CHAT_ENDPOINT="http://localhost:3001"
NEXT_PUBLIC_HOSTNAME="http://localhost:3003"
NEXT_PUBLIC_GOOGLE_OAUTH_CLIENT_KEY=""

These should be configured to match the endpoints where the backend services are available, ensuring the frontend can communicate with the backend properly.

Local Development

To run the application components separately for development purposes:

Run Backend

  cd backend && npm run dev

Run Frontend

  cd frontend && npm run dev

Build

To build all components for production deployment, follow these steps sequentially

cd shared
npm run build
cd ../backend
npm run build
cd ../frontend
npm run build

Running the Server

To start the server

Run Backend

  cd backend && npm run start

Run Chat Server (API Server)

  cd backend && npm run start:chat

Run Frontend

  cd frontend && npm run start