microsoftlearning / mslearn-ai-services Goto Github PK

I spent some time trying to figure out why the cloned repository did not contain a c# and python folder like 01 and 02.

I think that line should be removed for clarity.

Small change is required, should read "yourkey"

-H "Ocp-Apim-Subscription-Key: <your-endpoint>" to -H "Ocp-Apim-Subscription-Key: <your-key>"

This repo does not provide the link to https://microsoftlearning.github.io/mslearn-ai-services/

as shown in mslearn-ai-vision

As of fall 2023, new Personalizer resources can no longer be created, so this lab can't be done anymore.

In the page of this: https://microsoftlearning.github.io/mslearn-ai-services/Instructions/setup.html

It says about needs .NET 7.0 SDK installed:

This is problematic because as of today, 28th May 2024, .NET 7.0 is out of support:
(source: https://dotnet.microsoft.com/en-us/download/dotnet)

Could the setup.md and the actual lab instruction be updated to at least use .NET 8.0, the STS version?

https://github.com/MicrosoftLearning/mslearn-ai-services/blob/main/Instructions/Exercises/01-use-azure-ai-services.md

In lab no reminder to register the resource provider. But in fact by default you can't create AI service resources, you need to register the resource provider for cognitive services beforehand

Following the docs.

• Using python i am getting the error ("The Detect Language Operation under Text Analytics API (v3.1) is not supported with the current subscription key and pricing tier OpenAI.S0.)
• Using C# this one
  StatusCode: 401, ReasonPhrase: 'Access Denied', Version: 1.1, Content: System.Net.Http.HttpConnectionResponseContent, Headers:
  {
  apim-request-id: 52c36e73-48bb-48ef-b9a0-5e99a5d3e108
  apim-operation-name: Detect+Language
  Strict-Transport-Security: max-age=31536000; includeSubDomains; preload
  X-Content-Type-Options: nosniff
  WWW-Authenticate: AzureApiManagementKey realm="https://xxxxx.openai.azure.com/text/analytics",name="Ocp-Apim-Subscription-Key",type="header"
  Date: Wed, 06 Mar 2024 19:28:16 GMT
  Content-Length: 178
  Content-Type: application/json

In the Use an Azure AI Services Container lab, I created both an Azure container, and a local docker container.

I'm able to run the CURL command just fine with my localhost container instance, and it returns the languages of the content.

However, when I use the Azure container, I always receive this error: curl: (56) Recv failure: Connection was reset

I've created a second container, and it has the same issue. I've tried access with both IP and FQDN.

Is anyone else aware of this issue and how to resolve it?

In Lab 4, rest-test.sh has an incorrect curl syntax. It should be curl -X POST (as per lab guide), but is currently curl POST, which results in error: curl: (6) Could not resolve host: POST.

Should be Create: New Jupyter Notebook. Could lead to issues finding the command.

I'm on OSX using brew to install/maintain dotnet. i.e. brew install dotnet.
However, brew only has 8.0.1 and this code is still on version 7.0.0, which will be out of support by this May.
Since brew doesn't support version 7 (brew install [email protected]), I had to remove net8 and "hard" install net7.
Can this sample be upgraded to version 8 soon?
thanks,
Stuart Ngai.

Regarding Manage Azure AI Services Security.

Use the service principal in an application
...

1. In your terminal, switch to the C-Sharp or Python folder depending on your language preference. By running cd C-Sharp or cd Python. Then run cd keyvault_client.

For Python developers, the correct directory would be cd keyvault-client.

On the last section, for step 5, run the intended code python keyvault-client.py in the CLI. Once you input text the following error is logged: Parameter 'endpoint' must be a string.

In lab 02, the .env file has the variable COG_SERVICE_ENDPOINT, but the .py file is reading AI_SERVICE_ENDPOINT. As such it fails with a 'endpoint name must be a string' error. Changing the variable name to AI_SERVICE_ENDPOINT in .env fixes this.

LAB 2 - the C# code has "street.jpg" hardcoded in the code samples that remove the background. when you run the application it will always use that image when writing to background.png unless you change it