def test_save(trial_component_obj, sagemaker_boto_client):
trial_component_obj.display_name = str(uuid.uuid4())
trial_component_obj.status = api_types.TrialComponentStatus(
primary_status="InProgress", message="Message")
trial_component_obj.start_time = datetime.datetime.now(
datetime.timezone.utc) - datetime.timedelta(days=1)
trial_component_obj.end_time = datetime.datetime.now(datetime.timezone.utc)
trial_component_obj.parameters = {"foo": "bar", "whizz": 100.1}
trial_component_obj.input_artifacts = {
"snizz":
api_types.TrialComponentArtifact(value="s3:/foo/bar",
media_type="text/plain")
}
trial_component_obj.output_artifacts = {
"fly":
api_types.TrialComponentArtifact(value="s3:/sky/far",
media_type="away/tomorrow")
}
trial_component_obj.save()
loaded = trial_component.TrialComponent.load(
trial_component_name=trial_component_obj.trial_component_name,
sagemaker_boto_client=sagemaker_boto_client)
assert trial_component_obj.trial_component_name == loaded.trial_component_name
assert trial_component_obj.status == loaded.status
assert trial_component_obj.start_time - loaded.start_time < datetime.timedelta(
seconds=1)
assert trial_component_obj.end_time - loaded.end_time < datetime.timedelta(
seconds=1)
assert trial_component_obj.parameters == loaded.parameters
assert trial_component_obj.input_artifacts == loaded.input_artifacts
assert trial_component_obj.output_artifacts == loaded.output_artifacts
def test_save(trial_component_obj, sagemaker_boto_client):
trial_component_obj.display_name = str(uuid.uuid4())
trial_component_obj.status = api_types.TrialComponentStatus(primary_status='InProgress', message='Message')
trial_component_obj.start_time = datetime.datetime.now(datetime.timezone.utc) - datetime.timedelta(days=1)
trial_component_obj.end_time = datetime.datetime.now(datetime.timezone.utc)
trial_component_obj.parameters = {
'foo': 'bar',
'whizz': 100.1
}
trial_component_obj.input_artifacts = {
'snizz': api_types.TrialComponentArtifact(value='s3:/foo/bar', media_type='text/plain')
}
trial_component_obj.output_artifacts = {
'fly': api_types.TrialComponentArtifact(value='s3:/sky/far', media_type='away/tomorrow')
}
trial_component_obj.save()
loaded = trial_component.TrialComponent.load(trial_component_name = trial_component_obj.trial_component_name,
sagemaker_boto_client=sagemaker_boto_client)
assert trial_component_obj.trial_component_name == loaded.trial_component_name
assert trial_component_obj.status == loaded.status
assert trial_component_obj.start_time - loaded.start_time < datetime.timedelta(seconds=1)
assert trial_component_obj.end_time - loaded.end_time < datetime.timedelta(seconds=1)
assert trial_component_obj.parameters == loaded.parameters
assert trial_component_obj.input_artifacts == loaded.input_artifacts
assert trial_component_obj.output_artifacts == loaded.output_artifacts
def test_track_from_processing_job(sagemaker_boto_client, processing_job_name):
processing_job = sagemaker_boto_client.describe_processing_job(
ProcessingJobName=processing_job_name)
source_arn = processing_job['ProcessingJobArn']
wait_for_job(processing_job_name, sagemaker_boto_client)
trial_component_name = sagemaker_boto_client.list_trial_components(
SourceArn=source_arn
)['TrialComponentSummaries'][0]['TrialComponentName']
trial_component_obj = trial_component.TrialComponent.load(
trial_component_name=trial_component_name)
assert {
'p1': 1.0,
'SageMaker.InstanceType': 'ml.m5.large',
'SageMaker.InstanceCount': 1.0,
'SageMaker.VolumeSizeInGB': 10.0
} == trial_component_obj.parameters
image_uri = processing_job['AppSpecification']['ImageUri']
assert {
'SageMaker.ImageUri': api_types.TrialComponentArtifact(value=image_uri)
} == trial_component_obj.input_artifacts
assert not trial_component_obj.output_artifacts
assert not trial_component_obj.metrics
assert source_arn == trial_component_obj.source.source_arn
assert 'Completed' == trial_component_obj.status.primary_status
def test_load(sagemaker_boto_client):
now = datetime.datetime.now(datetime.timezone.utc)
sagemaker_boto_client.describe_trial_component.return_value = {
"TrialComponentArn": "A",
"TrialComponentName": "B",
"DisplayName": "C",
"Status": {"PrimaryStatus": "InProgress", "Message": "D"},
"Parameters": {"E": {"NumberValue": 1.0}, "F": {"StringValue": "G"}},
"InputArtifacts": {"H": {"Value": "s3://foo/bar", "MediaType": "text/plain"}},
"OutputArtifacts": {"I": {"Value": "s3://whizz/bang", "MediaType": "text/plain"}},
"Metrics": [
{
"MetricName": "J",
"Count": 1,
"Min": 1.0,
"Max": 2.0,
"Avg": 3.0,
"StdDev": 4.0,
"SourceArn": "K",
"Timestamp": now,
}
],
}
obj = trial_component.TrialComponent.load(trial_component_name="foo", sagemaker_boto_client=sagemaker_boto_client)
sagemaker_boto_client.describe_trial_component.assert_called_with(TrialComponentName="foo")
assert "A" == obj.trial_component_arn
assert "B" == obj.trial_component_name
assert "C" == obj.display_name
assert api_types.TrialComponentStatus(primary_status="InProgress", message="D") == obj.status
assert {"E": 1.0, "F": "G"} == obj.parameters
assert {"H": api_types.TrialComponentArtifact(value="s3://foo/bar", media_type="text/plain")}
assert {"I": api_types.TrialComponentArtifact(value="s3://whizz/bang", media_type="text/plain")}
assert [
api_types.TrialComponentMetricSummary(
metric_name="J", count=1, min=1.0, max=2.0, avg=3.0, std_dev=4.0, source_arn="K", timestamp=now
)
]
def log_output(self, name, value, media_type=None):
"""Record a single output artifact for this trial component.
Overwrites any previous value recorded for the specified output name.
Examples
.. code-block:: python
# log input dataset s3 location
my_tracker.log_output(name='prediction', value='s3://outputs/path')
Args:
name (str): The name of the output value.
value (str): The value.
media_type (str, optional): The MediaType (MIME type) of the value.
"""
self.trial_component.output_artifacts[
name] = api_types.TrialComponentArtifact(value,
media_type=media_type)
def log_artifact(self, file_path, name=None, media_type=None):
"""Upload a local file to s3 and store it as an artifact in this trial component.
Examples
.. code-block:: python
# log local artifact
my_tracker.log_artifact(file_path='/local/path/artifact.tar.gz')
Args:
file_path (str): The path of the local file to upload.
name (str, optional): The name of the artifact.
media_type (str, optional): The MediaType (MIME type) of the file. If not specified, this library
will attempt to infer the media type from the file extension of ``file_path``.
"""
media_type = media_type or _guess_media_type(file_path)
name = name or _resolve_artifact_name(file_path)
s3_uri = self._artifact_uploader.upload_artifact(file_path)
self.trial_component.output_artifacts[
name] = api_types.TrialComponentArtifact(value=s3_uri,
media_type=media_type)
def test_track_from_processing_job(sagemaker_boto_client, processing_job_name):
get_job = lambda: sagemaker_boto_client.describe_processing_job(
ProcessingJobName=processing_job_name)
processing_job = get_job()
source_arn = processing_job["ProcessingJobArn"]
wait_for_job(processing_job_name, get_job, "ProcessingJobStatus")
print(processing_job)
if "ProcessingStartTime" in processing_job:
start = to_seconds(processing_job["ProcessingStartTime"])
end = to_seconds(processing_job["ProcessingEndTime"])
print("Processing job took: {} seconds".format(end - start))
trial_component_name = sagemaker_boto_client.list_trial_components(
SourceArn=source_arn
)["TrialComponentSummaries"][0]["TrialComponentName"]
trial_component_obj = trial_component.TrialComponent.load(
trial_component_name=trial_component_name)
assert {
"SageMaker.InstanceType": "ml.m5.large",
"SageMaker.InstanceCount": 1.0,
"SageMaker.VolumeSizeInGB": 10.0,
} == trial_component_obj.parameters
image_uri = processing_job["AppSpecification"]["ImageUri"]
assert {
"SageMaker.ImageUri": api_types.TrialComponentArtifact(value=image_uri)
} == trial_component_obj.input_artifacts
assert not trial_component_obj.output_artifacts, "should be no output artifacts"
assert not trial_component_obj.metrics, "should be no metrics"
assert source_arn == trial_component_obj.source.source_arn, "source ARNs should match"
assert trial_component_obj.status.primary_status in [
"Completed", "Failed"
], "Status should be Completed or Failed"
def test_log_output(under_test):
under_test.log_output('foo', 'baz', 'text/text')
assert under_test.trial_component.output_artifacts == {
'foo':
api_types.TrialComponentArtifact(value='baz', media_type='text/text')
}
def test_load(sagemaker_boto_client):
now = datetime.datetime.now(datetime.timezone.utc)
sagemaker_boto_client.describe_trial_component.return_value = {
'TrialComponentArn':
'A',
'TrialComponentName':
'B',
'DisplayName':
'C',
'Status': {
'PrimaryStatus': 'InProgress',
'Message': 'D'
},
'Parameters': {
'E': {
'NumberValue': 1.0
},
'F': {
'StringValue': 'G'
}
},
'InputArtifacts': {
'H': {
'Value': 's3://foo/bar',
'MediaType': 'text/plain'
}
},
'OutputArtifacts': {
'I': {
'Value': 's3://whizz/bang',
'MediaType': 'text/plain'
}
},
'Metrics': [{
'MetricName': 'J',
'Count': 1,
'Min': 1.0,
'Max': 2.0,
'Avg': 3.0,
'StdDev': 4.0,
'SourceArn': 'K',
'Timestamp': now
}]
}
obj = trial_component.TrialComponent.load(
trial_component_name='foo',
sagemaker_boto_client=sagemaker_boto_client)
sagemaker_boto_client.describe_trial_component.assert_called_with(
TrialComponentName='foo')
assert 'A' == obj.trial_component_arn
assert 'B' == obj.trial_component_name
assert 'C' == obj.display_name
assert api_types.TrialComponentStatus(primary_status='InProgress',
message='D') == obj.status
assert {'E': 1.0, 'F': 'G'} == obj.parameters
assert {
'H':
api_types.TrialComponentArtifact(value='s3://foo/bar',
media_type='text/plain')
}
assert {
'I':
api_types.TrialComponentArtifact(value='s3://whizz/bang',
media_type='text/plain')
}
assert [
api_types.TrialComponentMetricSummary(metric_name='J',
count=1,
min=1.0,
max=2.0,
avg=3.0,
std_dev=4.0,
source_arn='K',
timestamp=now)
]
def test_log_output(under_test):
under_test.log_output("foo", "baz", "text/text")
assert under_test.trial_component.output_artifacts == {
"foo": api_types.TrialComponentArtifact(value="baz", media_type="text/text")
}