快速开始

1. 克隆本项目

git clone https://github.com/BSlience/dl-deploy.git

2. 安装依赖包

pip install -r requirements.txt

3. 下载部署中使用的模型，并确保放在 assets 目录下

wget http://aimaksen.bslience.cn/torch-bert-sentiment.bin -P assets/

4. 启动服务

uvicorn sentiment_analyzer.api:app

5. 测试服务

先安装 HTTPie，然后运行下面命令

# other console
http POST http://localhost:8000/predict text="OMG. I love how easy it is to stick to my schedule. Would recommend to everyone"

{
    "confidence": 0.9999295473098755,
    "probabilities": {
        "negative": 1.7753758584149182e-05,
        "neutral": 5.275019793771207e-05,
        "positive": 0.9999295473098755
    },
    "sentiment": "positive"
}