deployment issue in Azure.

Question

deployment issue in Azure.

santra 6

Hello, I am a beginner in Azure ML. I am have successfully deployed my NLP model in Azure for sentiment analysis. Then I go to the workspace and Azure Machine Learning services. When I go to the endpoints, I get a 'Test' option. But when I put a sentence for example 'I love to play football' in the test section the only things come up in the Test result
is {}. Can any one help.

1 answer

Your answer

Answer 1

romungi-MSFT 48,886 Microsoft Employee Moderator

@santra A null response in your endpoint result indicates that the scoring script used in your model deployment is not working as expected. Have you used any document to follow the steps and deploy the endpoint? Could you also lookup the logs of your endpoint to check more details?

santra 6 Reputation points

2021-09-20T11:20:06.07+00:00
I have no idea. The deployment is in a healthy state azure states. When I check in the Jupiter notebook,

import requests import json service = Webservice(workspace=ws, name="myserv") uri = service.scoring_uri requests.get("/score") headers = {"Content-Type": "application/json"} raw_data = { "query":"I love to play football", } data = json.dumps(raw_data) response = requests.post(uri, data=raw_data, headers=headers) print(response.json())

The error is saying JSON decode error"Expecting value: line 1 column 1 (char 0)" Also when I type

print(service.scoring_uri)

A url comes up.But when I click the url, the page opens up with "Unauthorized, no token matched". I think this may be the reason for the error.

romungi-MSFT 48,886 Microsoft Employee Moderator

@santra Couple of pointers to check. There should be a consume tab next to Test and you can use either C#/Python/R code snippets to call your scoring endpoint. This sample should look like something below for python.

data = {  
    "Inputs": {  
        "WebServiceInput0":  
        [  
            {  
                'FlipperLength': "181",  
                'BodyMass': "3750",  
            },  
        ],  
    },  
    "GlobalParameters": {  
    }  
}  
  
body = str.encode(json.dumps(data))  
  
url = 'http://<endpoint>.centralus.azurecontainer.io/score'  
api_key = '<endpoint_key>' # Replace this with the API key for the web service  
headers = {'Content-Type':'application/json', 'Authorization':('Bearer '+ api_key)}  
  
req = urllib.request.Request(url, body, headers)  
  
try:  
    response = urllib.request.urlopen(req)  
  
    result = response.read()  
    print(result)  
except urllib.error.HTTPError as error:  
    print("The request failed with status code: " + str(error.code))  
  
    # Print the headers - they include the requert ID and the timestamp, which are useful for debugging the failure  
    print(error.info())  
    print(json.loads(error.read().decode("utf8", 'ignore')))

You cannot directly access the endpoint without a token or key so the error with unauthorized token is valid.

santra 6 Reputation points

2021-09-23T13:36:12.03+00:00

No, I have sorted it out. There was a small problem (indentation) in the score.py file. Now everything works file.

Share via

deployment issue in Azure.

1 answer

Your answer