Files
qiming/qiming-mcp-proxy/fastembed/README.md
2026-06-01 13:03:20 +08:00

80 lines
1.5 KiB
Markdown

# FastEmbed
**[English](README.md)** | **[简体中文](README_zh-CN.md)**
---
# FastEmbed
Text embedding HTTP service using FastEmbed library for efficient text vectorization.
## Overview
`fastembed` is a high-performance text embedding service built with Rust, providing HTTP API for text vectorization using FastEmbed.
## Features
- **FastEmbed Integration**: Uses FastEmbed 5.0 with ONNX runtime
- **HTTP API**: RESTful API for text embedding
- **Concurrent Processing**: DashMap for efficient concurrent operations
- **OpenAPI Documentation**: Auto-generated API docs
- **Multiple Models**: Support for various embedding models
## Quick Start
### Installation
```bash
# Build from source
cargo build --release -p fastembed
# Binary location
ls target/release/fastembed
```
### Usage
```bash
# Start server (default port 8080)
fastembed server
# Specify custom port
fastembed server --port 8081
```
### API Usage
```bash
# Generate embeddings
curl -X POST http://localhost:8080/embed \
-H "Content-Type: application/json" \
-d '{
"texts": ["Hello world", "Fast embedding"],
"model": "BAAI/bge-small-en-v1.5"
}'
```
## Supported Models
- `BAAI/bge-small-en-v1.5` - Fast English model (384 dimensions)
- `BAAI/bge-base-en-v1.5` - Balanced English model (768 dimensions)
- `BAAI/bge-large-en-v1.5` - High-quality English model (1024 dimensions)
## Development
```bash
# Build
cargo build -p fastembed
# Test
cargo test -p fastembed
```
## License
MIT OR Apache-2.0
## Contributing
Issues and Pull Requests are welcome!