metadata
title: π§ intfloat/e5-small Embedding API Server
emoji: π§©
colorFrom: blue
colorTo: purple
sdk: docker
app_file: app.py
pinned: false
π§ Local Embedding API Server (Multilingual)
A lightweight embedding generation server using FastAPI + SentenceTransformer, supporting Chinese, English, Malay and more β powered by BAAI/bge-m3.
π Features
/embed: generate embedding for a single string/embed/batch: generate embedding for multiple strings- Support for GPU acceleration (CUDA) via Docker
- Docker-ready deployment
π§ Requirements
- Docker (with NVIDIA container runtime)
- Optional: Python 3.10+ for local dev
βοΈ Run with Docker (GPU)
docker build -t embed-server .
docker run --gpus all --env-file .env -p 8000:8000 embed-server
π¦ API Usage
POST /embed
{
"input": "Saya suka makan nasi lemak"
}
POST /embed/batch
{
"inputs": ["I like cats", "ζεζ¬’η«", "Saya suka kucing"]
}
π€ Response
{
"embedding": [...],
"dim": 1024,
"model": "altaidevorg/BGE-M3-Distill-8L"
}
βοΈ License
MIT