Embedding FastAPI Server

This is a FastAPI server for generating sentence embeddings using Transformer models like BAAI/bge-base-en-v1.5 and sentence-transformers/all-mpnet-base-v2.

Features

• Generate embeddings for text input using HTTP API
• Support for multiple models with different endpoints
• Async generation using run_in_threadpool
• Dockerized for easy deployment

Usage

Local Development

# Install dependencies
pip install -r requirements.txt

# Start server
uvicorn app:app

The server will be running at http://127.0.0.1:8000.

You can make requests like:

curl -X POST "http://127.0.0.1:8000/v1/embeddings/baai" -H  "accept: application/json" -H  "Content-Type: application/json" -d "{\"input\": \"Hello world\"}"

Docker

# Build image
docker build -t embedding-server .

# Start container
docker run -p 80:80 embedding-server

The server will be running at http://localhost

Models

Currently supported models:

• baai: BAAI/bge-base-en-v1.5
• st: sentence-transformers/all-mpnet-base-v2

Add a new model by:

1. Loading model in load_model in app.py
2. Adding to MODELS dictionary
3. Creating a new endpoint

License

MIT