Predictions
Model predictions may be deployed by calling Model.predict_batches
or Model.predict
directly.
m = db.load('model', 'my-model')
# *args, **kwargs depend on model implementation
results = m.predict(*args, **kwargs)
An alternative is to construct a prediction "query" as follows:
# *args, **kwargs depend on model implementation
db['my-model'].predict(*args, **kwargs).execute()
The results should be the same for both versions.