A lightweight FastAPI server for generating 384-dimensional dense vector embeddings from text using the MiniLM-L6 model (all-MiniLM-L6-v2). Useful for semantic search, text similarity, and other NLP tasks.
- Simple REST API for text embedding
- Fast inference with Sentence Transformers
- Docker and direct Python support
- Returns embeddings as JSON arrays
- Handles invalid input with clear error messages
- Build the Docker image:
docker build -t balena-embed . - Run the server:
docker run -p 6700:6700 balena-embed
- Install Python 3.11 (or 3.11 slim) and pip if not already installed.
- Install dependencies:
pip install -r requirements.txt
- Start the server:
python embed_server.py
Send a POST request to http://localhost:6700/embed with JSON body:
{
"text": "your text here"
}The response will be a JSON object with a 384-dimensional embedding vector:
{
"embedding": [0.123, ...]
}- If the
textfield is missing or not a string, the server returns a 422 error with details. - If the model fails to generate an embedding, a 500 error is returned.
- Python 3.11 or 3.11 slim
- pip (for direct install)
- Docker (for containerized install)
Contributions are welcome! Please open an issue or submit a pull request.
This project is licensed under the GPL 3.0 License. See the LICENSE file for details.