mModelShare

---

This open source mModelShare microservice created by the mimik team is an example cross-platform solution for sharing machine learning models across devices. More specifically, this edgeSDK microservice has the following functionality:

1. Uploading the information about the machine learning model on your device to the mModelShare's local database.
2. Retrieving the information regarding specific machine learning models from local database.
3. Retrieving machine learning models from other devices.

Build Process

---

The build script default.yml is specified under config directory.

1. Install dependencies: npm install
2. Run the build script: npm run build
3. Package to container: npm run package

Deployment

---

For mobile application development, deployment is programmatically by Android or iOS Wrappers, learn more about it:

• Android: Link
• iOS: Link

For microservice development, things you will need:

• edgeSDK running on the deployment targeted device.
• Obtained edge Acess Token and associate the device from edgeSDK OAuth Tool.
• Run the following commands under the same directory of your containerized microservice file:

curl -i -H "Authorization: Bearer <edge Access Token>" -F "[email protected]" http://localhost:8083/mcm/v1/images

• To run the microservice after successful deployment, with environment variables:

curl -i -H "Authorization: Bearer <edge Access Token>" -d '{"name": "model-v1", "image": "model-v1", "env": {"MCM.BASE_API_PATH": "/modelshare/v1", "uMDS": "http://localhost:8083/mds/v1"} }' http://localhost:8083/mcm/v1/containers

• For more information and explanation, you can visit our mCM container management API references and general guide on packaing, deployment, and exporting microservice.