I have trained my custom text classification model on azure language studio and also deployment it but when I want to send curl request to it from cmd then it gives below error. I have used get prediction URL method to query my model. How can I solve this error? {"error":{"code":"InvalidRequest","message":"Invalid Request.","innererror":{"code":"EmptyRequest","message":"Request body must be present."}}}

Muhammad Qasim I think your request is missing the body section or is incorrectly formatted. Are you using a custom question answering KB deployment prediction URL? In that case, you can use the Get Prediction URL button to copy the curl request and then use the same. For example, I have used this option to copy the curl request and run the same with minimal parameters. curl -X POST "https://<your_resource>.cognitiveservices.azure.com/language/:query-knowledgebases?projectName=<project_name>&api-version=2021-10-01&deploymentName=production" -H "Ocp-Apim-Subscription-Key: <your_key>" -H "Content-Type: application/json" -d "{\"top\":3,\"question\":\"This is test\",\"includeUnstructuredSources\":true,\"confidenceScoreThreshold\":\"0.4\",\"answerSpanRequest\":{\"enable\":true,\"topAnswersWithSpan\":1}}}" If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.

I want to use my deployed azure language studio model in a website.

Accepted answer

romungi-MSFT 41,961 Reputation points Microsoft Employee

2023-03-15T10:20:27.3666667+00:00

Muhammad Qasim I think your request is missing the body section or is incorrectly formatted. Are you using a custom question answering KB deployment prediction URL? In that case, you can use the Get Prediction URL button to copy the curl request and then use the same.

For example, I have used this option to copy the curl request and run the same with minimal parameters.

curl -X POST "https://<your_resource>.cognitiveservices.azure.com/language/:query-knowledgebases?projectName=<project_name>&api-version=2021-10-01&deploymentName=production" -H "Ocp-Apim-Subscription-Key: <your_key>" -H "Content-Type: application/json" -d "{\"top\":3,\"question\":\"This is test\",\"includeUnstructuredSources\":true,\"confidenceScoreThreshold\":\"0.4\",\"answerSpanRequest\":{\"enable\":true,\"topAnswersWithSpan\":1}}}"

If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.
Please sign in to rate this answer.

1 person found this answer helpful.
Muhammad Qasim 41 Reputation points

2023-03-15T11:16:09.4733333+00:00

I also tried get prediction url method but when I paste in in my cmd and press enter then gives error as well. I also specified the text and language but gives error
....ation","analysisInput":{"documents":[{"id":"document_CustomSingleLabelClassification","text":"YOUR_DOCUMENT_HERE","language":"YOUR_DOCUMENT_LANGUAGE_HERE"}]}}"

here I specified language and given text also but error

will I get 200 in response when the request is successful?

i tried it in flask and output was 200 but how can I display response from json()

romungi-MSFT 41,961 Reputation points Microsoft Employee

2023-03-15T12:55:48.1366667+00:00

Have you tried to use the REST API reference to construct your request and then format it to use through cmd via curl? I usually try to use this to build the request body or just use postman for REST API for ease of use.

With respect to the flask application that you are referring. How is this setup to call the API? If a 202 is received, then you should be able to print the operation-location header of the result & try a GET request on this URL. This will should get the response body when the job is complete.

Muhammad Qasim 41 Reputation points

2023-03-15T13:05:58.6866667+00:00

I used this module to deploy flask application https://learn.microsoft.com/en-us/training/modules/python-flask-build-ai-web-app/

how can i use GET request to get output from model

romungi-MSFT 41,961 Reputation points Microsoft Employee

2023-03-15T13:39:33.0733333+00:00

This seems to be using translation endpoint in which result is directly available from the POST request response. In case of async requests like classification you will get an operation-location header in the response which basically points to the location of your result on Azure. This URL should be used in GET request with the same subscription key header used in your original request. If you are storing the response in a variable, the header can be printed with variable.headers which should contain the operation-location to be used in GET request.

Muhammad Qasim 41 Reputation points

2023-03-15T14:14:06.6833333+00:00

I am using custom text classification in language studio and I used response=requests.post(url,headers,json) Now how i get result that my input is offensive or not offensive. I used single label classification and my Input will either offensive or not offensive.

romungi-MSFT 41,961 Reputation points Microsoft Employee

2023-03-15T15:04:06.01+00:00

Using the studio, you are able to use the UI to handle the process of submitting the job and getting the response but if you want to use the API, you will have to first use the runtime API to classify text. Please see the next steps link at the bottom of this page.

The runtime API is an async API which takes your document or text along with the custom classification project that you have developed from the studio. A sample input will have the body with your project details.

{ "displayName": "Classifying documents", "analysisInput": { "documents": [ { "id": "1", "language": "{LANGUAGE-CODE}", "text": "Text1" }, { "id": "2", "language": "{LANGUAGE-CODE}", "text": "Text2" } ] }, "tasks": [ { "kind": "CustomSingleLabelClassification", "taskName": "Single Classification Label", "parameters": { "projectName": "{PROJECT-NAME}", "deploymentName": "{DEPLOYMENT-NAME}" } } ] }

Once this request is submitted through POST, you can check the status of response i.e a 202 accepted if successful. This will provide the operation-location id or job id which will contain the actual result of your text classification. It will be of the following format:

{ENDPOINT}/language/analyze-text/jobs/{JOB-ID}?api-version={API-VERSION}

Now run a GET request on this URL with the same subscription key as header for authentication and the resulting JSON contains the result or category of your document or text. A sample response snippet is below.

"items": [ { "kind": "customSingleClassificationTasks", "taskName": "Classify documents", "lastUpdateDateTime": "2020-10-01T15:01:03Z", "status": "succeeded", "results": { "documents": [ { "id": "{DOC-ID}", "class": [ { "category": "Class_1", "confidenceScore": 0.0551877357 } ], "warnings": [] } ], "errors": [], "modelVersion": "2020-04-01" } } ]

I hope this helps!!

Muhammad Qasim 41 Reputation points

2023-03-15T16:02:57.57+00:00

Where can I use this post and get request? In postman?

romungi-MSFT 41,961 Reputation points Microsoft Employee

2023-03-15T17:41:53.7633333+00:00

I would recommend using postman if you are checking the same for the first time. You can then implement this using any language to submit the request and then poll to get the result of the job.

Muhammad Qasim 41 Reputation points

2023-03-15T18:05:40.76+00:00

In postman when i send post method then it gives 202 as processing the request and when send get (as described by you above) then gives 404 error

romungi-MSFT 41,961 Reputation points Microsoft Employee

2023-03-16T04:15:04.8333333+00:00

Any status other than 200 is possibly the job failed which could be due to input or a service issue. Do you see any other details in the JSON about the failure? Because, the request is async the job could fail if the inputs are not correct or deployed endpoint failed to process the job. Does the same input work with studio?

Sample response:

Muhammad Qasim 41 Reputation points

2023-03-16T08:25:47.87+00:00

Kindly tell me that how can I check that my resource is getting api calls or not in postman. I checked the metrics for my resource and shows that some calls are successful but I didn't got output. Can you come on Skype for video call?

romungi-MSFT 41,961 Reputation points Microsoft Employee

2023-03-16T08:54:49.4666667+00:00

Muhammad Qasim The metrics blade from azure portal should show you the requests that are successful and ones that are not. Looking up the job status should provide you more details on why the job failed.

You could raise a support case from Azure portal for 1:1 help as the service team would be able to advise on what could be wrong with the calls as they would be able to lookup the request ids internally and your project and deployment configurations.

Muhammad Qasim 41 Reputation points

2023-03-16T14:17:26.6833333+00:00

i used the above get and post methods , what is the issue?

romungi-MSFT 41,961 Reputation points Microsoft Employee

2023-03-16T14:39:31.69+00:00

From the first screen shot I do not see the Job id in the request. The placeholder value is not updated.

The second screen shot has 202 accepted. Try checking the Headers(8) tab to get the operation-location header and use that URL in the GET request. This should have the job id in the URL. Thanks!!

Muhammad Qasim 41 Reputation points

2023-03-16T14:46:38.71+00:00

ye got it thanks bro. love you🥰🥰🥰🥰🥰🥰🥰

Muhammad Qasim 41 Reputation points

2023-03-16T14:48:12.57+00:00

now how can i use it in flask?

romungi-MSFT 41,961 Reputation points Microsoft Employee

2023-03-17T05:29:43.6733333+00:00

If you want to use the REST API with your application, you have to follow similar pattern to POST a call and then use the operation-location URL with GET. Since this is async operation you have to ensure a retry mechanism is in place till your results are available after reading the response. I have not worked on flask before, but you can achieve this using any library that can simply handle HTTP calls.

It would be great if you could accept my response as answer or rate the response as helpful. Thanks!!

Muhammad Qasim 41 Reputation points

2023-03-17T05:32:56.3+00:00

Thanks a lot finally deployed I am rating your answers
Sign in to comment

I want to use my deployed azure language studio model in a website.

0 additional answers