Repository navigation
model.predict_proba is not working in scoring script for classification #1950
Description
Activity
I noticed the scoring script in the example uses predict_proba, but my deployed endpoint returns only class labels.
Could you please clarify the exact location of the scoring_script.py used during deployment?
I want to confirm that the correct script version is being packaged inside the code_configuration directory
This is run function. But My output of test data is class only not probabilities.
@input_schema('data', PandasParameterType(input_sample))
@output_schema(NumpyParameterType(output_sample))
def run(data):
result=model.predict_proba(data)
return result.tolist()
My deployement Function for reference
deployment = ManagedOnlineDeployment(
name=deployment_name,
endpoint_name=online_endpoint_name,
model=registered_model.id,
instance_type="Standard_F4s_v2",
instance_count=1,
code_configuration=CodeConfiguration(
code="./artifact_downloads/outputs",
scoring_script="scoring_script.py",
),
liveness_probe=ProbeSettings(
failure_threshold=30,
success_threshold=1,
timeout=2,
period=10,
initial_delay=2000,
),
readiness_probe=ProbeSettings(
failure_threshold=10,
success_threshold=1,
timeout=10,
period=10,
initial_delay=2000,
),
)