ServiceDefinition

Class ServiceDefinition

azureml.deploy.operationalization.ServiceDefinition(name, op)

Bases: azureml.deploy.operationalization.OperationalizationDefinition

Service class defining a standard service's properties for publishing.

alias(alias)

Set the optional service function name alias to use in order to consume the service.

Example:

service = client.service('score-service').alias('score').deploy()

# `score()` is the function that will call the `score-service`
result = service.score()

Arguments

alias

The service function name alias to use in order to consume the service.

Returns

Self OperationalizationDefinition for fluent API.

artifact

artifact(artifact)

Define a service's optional supported file artifact by name. A convenience to calling .artifacts(['file.png']) with a list of one.

Arguments

artifact

A single file artifact by name.

Returns

Self OperationalizationDefinition for fluent API chaining.

artifacts

artifacts(artifacts)

Defines a service's optional supported file artifacts by name.

Arguments

artifacts

A list of file artifacts by name.

Returns

Self OperationalizationDefinition for fluent API chaining.

code_fn

code_fn(code, init=None)

Set the service consume function as a function.

Example:

def init():
    pass

def score(df):
    pass

.code_fn(score, init)

Arguments

code

A function handle as a reference to run python code.

init

An optional function handle as a reference to initialize the service.

Returns

Self OperationalizationDefinition for fluent API chaining.

code_str

code_str(code, init=None)

Set the service consume function as a block of python code as a str.

init = 'import pandas as pd'
code = 'print(pd)'

.code_str(code, init)

Arguments

code

A block of python code as a str.

init

An optional block of python code as a str to initialize the service.

Returns

A ServiceDefinition for fluent API chaining.

deploy

deploy()

Bundle up the definition properties and publish the service.

Returns

A new instance of Service representing the service deployed.

description

description(description)

Set the service's optional description.

Arguments

description

The description of the service.

Returns

Self OperationalizationDefinition for fluent API.

inputs

inputs(**inputs)

Defines a service's optional supported inputs by name and type.

Example:

.inputs(a=float, b=int, c=str, d=bool, e='pandas.DataFrame')

Arguments

inputs

The inputs by name and type.

Returns

Self OperationalizationDefinition for fluent API chaining.

models

models(**models)

Include any model(s) used for this service.

Example:

cars_model = rx_lin_mod(formula="am ~ hp + wt",data=mtcars)

.models(cars_model=cars_model)

Arguments

models

Any models by name and value.

Returns

Self OperationalizationDefinition for fluent API chaining.

objects

objects(**objects)

Include any object(s) used for this service.

Example:

x = 5
y = 'hello'

.objects(x=x, y=y)

Arguments

objects

Any objects by name and value.

Returns

Self OperationalizationDefinition for fluent API chaining.

outputs

outputs(**outputs)

Defines a service's optional supported outputs by name and type.

Example:

.outputs(a=float, b=int, c=str, d=bool, e='pandas.DataFrame')

Arguments

outputs

The outputs by name and type.

Returns

Self OperationalizationDefinition for fluent API chaining.

redeploy

redeploy(force=False)

Bundle up the definition properties and update the service.

Returns

A new instance of Service representing the service deployed.

version

version(version)

Set the service's optional version.

Arguments

version

The version of the service.

Returns

Self OperationalizationDefinition for fluent API.