Deploying autoML model locally

Emanuel 0

Hello everyone, I'm fairly new to Azure, and I'm currently facing an issue that I can't seem to resolve. I've successfully trained some very promising models using Azure AutoML and now I want to deploy them locally. I used simple CSV files as datasets (using Azure ML v1 APIs). When I try to download the corresponding model and insert it into the Conda environment (set up as specified by Azure), I always get the error message:

---------------------------------------------------------------------------
ModuleNotFoundError                       Traceback (most recent call last)
Cell In[1], line 12
      9 import pandas as pd
     10 import joblib
---> 12 import azureml.automl.core
     13 from azureml.automl.core.shared import logging_utilities, log_server
     14 from azureml.telemetry import INSTRUMENTATION_KEY

ModuleNotFoundError: No module named 'azureml'

However, I don't believe it's related to packages. When I attempt to deploy the models via a real-time endpoint in Azure and test them using the "Test" tab after successful implementation, I receive the following error message:

Failed to test endpoint
{"message":"An unexpected error occurred in scoring script. Check the logs for more info."}

Am I overlooking something? Appreciate any help. Thanks, Emanuel

Ramr-msft 17,741 Reputation points

2024-01-22T16:38:38.22+00:00

Thanks for the question, Local Deployment Issue: The error message ModuleNotFoundError: No module named 'azureml' suggests that the ‘azureml’ package is not installed in your current Python environment. You can install it using pip with the command pip install azureml-sdk[automl]. If you’re using a Conda environment, make sure to activate the correct environment before installing the package. If the issue persists, you might want to check your Python path and ensure it’s pointing to the correct environment where ‘azureml’ is installed. Real-time Endpoint Deployment Issue: The error message {"message":"An unexpected error occurred in scoring script. Check the logs for more info."} is quite generic and could be caused by various issues. It’s recommended to check the logs for more detailed information. If the issue persists, you might want to try local deployment for testing and debugging before deploying to the cloud.
Deleted

This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.
Deleted

This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.
Deleted

This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.

Emanuel 0

In conda the error still occurs after trying **pip install azureml-sdk[automl]. **In my opinion the right environment is selected. Interestingly the scoring script is working in VS, but there I've got the following error after running run(input_sample) :

Invalid input data type to parse. Expected: <class 'dict'> but got <class 'inference_schema.parameter_types.pandas_parameter_type.PandasParameterType'>
Stapelüberwachung:
 >  File "C:\Users\Emanuel\Downloads\weightedfinalmodel\scoring_file_v_2_0_0.py", line 59, in <module> (Current frame)
 >    run(data_sample)
"inference_schema.schema_decorators" geladen
"__main__" geladen
"runpy" geladen
Das Programm "python.exe" wurde mit Code 4294967295 (0xffffffff) beendet.

Ramr-msft 17,741

Thanks for the details and sharing the code. The run function is expecting a dictionary as input, but it’s receiving a PandasParameterType object. The PandasParameterType is a wrapper around a pandas DataFrame that’s used for schema generation and validation in Azure Machine Learning. When you call the run function, you should pass in a dictionary where the 'data' key corresponds to a pandas DataFrame. However, in your input_sample, you’re wrapping the DataFrame in a PandasParameterType, which is causing the error. Here’s how you can modify the input_sample:

input_sample = {'data': pd.DataFrame({"SIO2": pd.Series([74.0], dtype="float32"), 
                                      "AL2O3": pd.Series([0.0], dtype="float32"), 
                                      "B2O3": pd.Series([0.0], dtype="float32"), 
                                      "CAO": pd.Series([12.0], dtype="float32"), 
                                      "K2O": pd.Series([0.0], dtype="float32"), 
                                      "NA2O": pd.Series([14.0], dtype="float32"), 
                                      "PBO": pd.Series([0.0], dtype="float32"), 
                                      "Li2O": pd.Series([0.0], dtype="float32"), 
                                      "MgO": pd.Series([0.0], dtype="float32"), 
                                      "SRO": pd.Series([0.0], dtype="float32"), 
                                      "BAO": pd.Series([0.0], dtype="float32"), 
                                      "ZNO": pd.Series([0.0], dtype="float32"), 
                                      "P2O5": pd.Series([0.0], dtype="float32"), 
                                      "ZRO2": pd.Series([0.0], dtype="float32"), 
                                      "TIO2": pd.Series([0.0], dtype="float32"), 
                                      "Bi2O3": pd.Series([0.0], dtype="float32")})}

Share via

Deploying autoML model locally

Your answer