---
title: BGE-M3 Embedding Service
emoji: 🚀
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 5.35.0
app_file: app.py
pinned: false
license: apache-2.0
---

# BGE-M3 Embedding Service

A high-performance embedding service powered by BAAI/bge-m3 model, supporting 100+ languages including Thai and English.

## Features

- 🌍 **Multilingual Support**: Works with 100+ languages
- ⚡ **Fast Inference**: Optimized for quick embedding generation  
- 🎯 **High Quality**: State-of-the-art embedding quality
- 🔧 **API Ready**: Easy integration with your applications
- 📊 **Similarity Computation**: Built-in cosine similarity calculation

## Usage

### Web Interface
Use the Gradio interface to generate embeddings interactively.

### API Integration
```python
import requests
import json

API_URL = "https://your-username-bge-m3-embedding.hf.space/api/predict"

data = {
    "data": [
        "สวัสดีครับ\nHello world",  # Input texts
        "embeddings"  # Operation
    ]
}

response = requests.post(API_URL, json=data)
result = response.json()
embeddings = json.loads(result["data"][0])["embeddings"]
```

## Model Details

- **Model**: BAAI/bge-m3
- **Embedding Dimension**: 1024
- **Input Length**: Up to 8192 tokens
- **Output**: Normalized embeddings

## Applications

- Semantic search systems
- RAG (Retrieval-Augmented Generation)
- Text similarity analysis
- Document clustering
- Multilingual information retrieval

## License

Apache 2.0