Embedding Service

High-performance text embedding service with caching

Category: embedding
Version: 1.0.0
Deployment formats: Docker Compose, Helm (Kubernetes), Terraform (local Docker), Terraform (AWS)
Maintainer: Hugging Face
Tags: Embeddings, Hugging Face, Semantic Search, API, FastAPI

Overview

High-performance text embedding service using Hugging Face models. Supports batch processing, caching, and horizontal scaling. Perfect for semantic search and similarity matching.

Requirements

Docker installed
4GB RAM minimum
First run downloads model (130MB - 1.3GB depending on model)

Quick start

Quick Start (Docker)

Start the service:

   docker-compose up -d

Wait for model download (first run: ~1.3GB for bge-large)

Test the API:

   curl http://localhost:8080/embed \
     -X POST \
     -H "Content-Type: application/json" \
     -d '{"inputs": "What is machine learning?"}'

Batch embeddings:

   curl http://localhost:8080/embed \
     -X POST \
     -H "Content-Type: application/json" \
     -d '{"inputs": ["First text", "Second text"]}'

Model Options:

BAAI/bge-small-en-v1.5 - Fast (130MB)
BAAI/bge-base-en-v1.5 - Balanced (440MB)
BAAI/bge-large-en-v1.5 - Best quality (1.3GB)

Environment variables

MODEL_ID – Hugging Face model ID for embeddings

docker-compose.yml

version: "3.8"

services:
  embeddings:
    image: ghcr.io/huggingface/text-embeddings-inference:cpu-1.5
    container_name: text-embeddings
    ports:
      - "8080:80"
    environment:
      - MODEL_ID=BAAI/bge-large-en-v1.5
    volumes:
      - models:/data
    restart: unless-stopped
    # For GPU support, use this image instead:
    # image: ghcr.io/huggingface/text-embeddings-inference:1.5

volumes:
  models: