You need 3 things to create a new deployment:
An artifact file. It will be copied to
/home/mlstudio/models for later usage and reproducablity.
A valid requirements.txt file. It will be validated on submission to make sure everything there works.
A python class for running the model. A python class with your inference code.
class UserModel(object):def __init__(self):# Load your model herepassdef predict(self, data):# Your predication code goes herepass
json you would sent when calling the generated endpoint.
Once you have the above files, open ML Studio and go to
To add a new deployment, press
New Deployment where you can add your model name, version and short description.
Then you will be prompted to select the files from the start of this page. And choose your prefered deployment infrastructure.
Finally, just wait a few minutes untill it is ready.