secretflow-serving-lib 0.5.0b0

Serving is a subproject of Secretflow that implements model serving capabilities.

SecretFlow-Serving

SecretFlow-Serving is a serving system for privacy-preserving machine learning models.

Serve a model

# Download the Secretflow Serving Docker image
docker pull secretflow/serving-anolis8:latest

# Start Secretflow Serving container and open the REST API port
cd examples

docker-compose up -d

# Query the model using the predict API
curl --location 'http://127.0.0.1:9010/PredictionService/Predict' \
    --header 'Content-Type: application/json' \
    --data '{
        "service_spec": {
            "id": "test_service_id"
        },
        "fs_params": {
            "alice": {
                "query_datas": [
                    "a"
                ]
            },
            "bob": {
                "query_datas": [
                    "a"
                ]
            }
        }
    }'

Contribution Guidelines

If you would like to contribute to SecretFlow-Serving, please check Contribution guidelines.

This documentation also contains instructions for build and testing.