Gain insights into Azure basics, build ML/deep learning pipelines, manage deployments, and analyze models ethically with responsible AI, ensuring fairness and bias mitigation.

“Build, Deploy MachineDeep Learning Models using Azure CLI”.png

azure.tar.gz

UserName

Location

ResourceGroup

Workspace

TenantId

AppId

Password

SubscriptionId

Compute

StorageKey

StorageAccount

OnlineEndpoint

BatchEndpoint

AppId_US

Password_US

RG1_US

WS1_US

Compute_US

scoring_uri

api_key

VirtualMachine

run_id

PrivateWorkspaceCreation

MLExtension

DefaultWorkspace

ResourceGroupCreation

ServicePrincipal

VMCreation

ComputeCreate

ResourceGroup Update

ComputeDelete

Demo-run-command-copy

Endpoint-Get Credentials

ComputeCreate-yml

EnvCreate-Docker

EnvCreate

EnvList

DataCreate-Simple

Storage Account Creation

Storage Account ContainerCreation

Storage Blob Upload

Storage Blob Download

Storage Blob Delete

JobCreate

Job-run-SPA

DataCreate-yml

VMList

Model-Create-SPA

Component-Create-SPA

endpoint-Create-SPA

ComputeDelete-copy

endpoint-uai-create-SPA

endpoint-batch-create-SPA

deployment-batch-create-SPA

deployment-online-create-SPA

endpoint-local-Create-SPA

RunCommand

VM-Create-SPA

PrivateWorkspaceCreation-copy

endpoint-Create-Local-SPA

deployment-batch-invoke

Run-command-SPA

In this course, you will start your journey by gaining a comprehensive understanding of the basics of Azure, including creating and managing Azure resources, and setting up the Azure CLI environment. 

Next, you will learn how to build Azure Machine Learning pipelines from scratch. Then, you’ll delve into deep learning and distributed deep learning pipelines. You’ll learn how to manage the deployment and scheduling of these models. You’ll also cover the complete model management strategies. Finally, the course will cover model analysis using responsible AI, teaching how to identify and mitigate potential biases in models and ensuring that models are ethical and fair. You’ll also learn how to analyze the models and identify potential areas for improvement.

By the end of this course, you’ll have a comprehensive understanding of Azure Machine Learning, including how to build complex pipelines, deploy models using online/batch methods, and manage and analyze models using tools like MLflow and responsible AI.

Designing Machine/Deep Learning Models Using Azure CLI

Which type of deployment mode should be used if we want to integrate ML models in real-time?

Which type of deployment mode should be used if we want to integrate ML models in real-time?

`az ml online-endpoint set --traffic "blue-depl=95 green-depl=5"`

`az ml online-endpoint update --traffic "blue-depl=95 green-depl=5"`

`az ml online-endpoint update --traffic "blue-depl=5 green-depl=95"`

`az ml online-endpoint --traffic "blue-depl=95 green-depl=5"`

If we want to set the `blue-depl` to 95% and `green-depl` to 5%, which command is correct?

If we want to set the <code>blue-depl</code> to 95% and <code>green-depl</code> to 5%, which command is correct?

```yaml
$schema: https://azuremlschemas.azureedge.net/latest/batchDeployment.schema.json

name: diabetes-depl

endpoint_name: batch-endpt

model: 
  path: ./Diabetes-model

compute: azureml:ComputeEdu
```

```yaml
$schema: https://azuremlschemas.azureedge.net/latest/OnlineDeployment.schema.json

name: diabetes-depl

endpoint_name: batch-endpt

model: 
  path: ./Diabetes-model

compute: azureml:ComputeEdu
```

```yaml
$schema: https://azuremlschemas.azureedge.net/latest/batchDeployment.schema.json

name: diabetes-depl

endpoint_name: batch-endpt

model: 
  path: ./Diabetes-model

compute: ComputeEdu
```

```yaml
$schema: https://azuremlschemas.azureedge.net/latest/batchDeployment.schema.json

name: diabetes-depl1

endpoint_name: batch-endpt

model: 
  path: ./Diabetes-model

compute: Standard_DS3_v2
```

Which of these YAML configurations are the right configurations for batch deployment?

Which of these YAML configurations are the right configurations for batch deployment?

```
{"input_data": {
    "columns": [
      "sepal_length",
      "sepal_width",
      "petal_length"
    ],
    "data": [
      [ 1.0,2.0,5.0,6.0],
      [ 10.0,2.0,9.0,8.0]
    ],
    "index": [0,1]
  }}
```

```
{"input_data": {
    "columns": [
      "sepal_length",
      "sepal_width",
      "petal_length",
      "petal_width"
    ],
    "data": [
      [ 1.0,2.0,5.0,6.0],
      [ 10.0,2.0,9.0]
    ],
    "index": [0,1]
  }}
```

```
{"input_data": {
    "columns": [
      "sepal_length",
      "sepal_width",
      "petal_length",
      "petal_width"
    ],
    "data": [
      [ 1.0,2.0,5.0,6.0],
      [ 10.0,2.0,9.0,8.0]
    ],
    "index": [0,1]
  }}
```

```
{"input_data": {
    "columns": [
      "sepal_length",
      "sepal_width",
      "petal_length",
      "petal_width"
    ],
    "data": [
      [ 1.0 2.0 5.0 6.0],
      [ 10.0 2.0 9.0 6.5]
    ],
    "index": [0,1]
  }}
```

Which request file is a correctly formatted file for sending the request to online deployment?

Which request file is a correctly formatted file for sending the request to online deployment?

The deployment name `green-depl-2` is wrong. It should not contain digits.

`Azureml:Model:2` is not present in Azure.

The deployment name is mismatched in the command vs. deployment file, i.e., `green-depl` vs. `geen-depl-2`.

Suppose a user has initiated a deployment using the following YAML file. The deployment failed with a `BadArgument` error. What is likely the cause?

Deployment YAML:
```yaml
$schema: https://azuremlschemas.azureedge.net/latest/managedOnlineDeployment.schema.json
name: green-depl-2
endpoint_name: online-endpt
model:
  path: Azureml:Model:2
code_configuration:
  code: model/onlinescoring/
  scoring_script: score.py
environment: 
  conda_file: model/environment/conda.yml
  image: mcr.microsoft.com/azureml/openmpi3.1.2-ubuntu18.04:20210727.v1
instance_type: Standard_DS2_v2
instance_count: 1
```
Deployment is triggered using the following command:
```
az deployment create --name geen-depl-2
```

Suppose a user has initiated a deployment using the following YAML file. The deployment failed with a <code>BadArgument</code> error. What is likely the cause?
Deployment YAML:
<pre><code class="language-yaml">$schema: https://azuremlschemas.azureedge.net/latest/managedOnlineDeployment.schema.json
name: green-depl-2
endpoint_name: online-endpt
model:
 path: Azureml:Model:2
code_configuration:
 code: model/onlinescoring/
 scoring_script: score.py
environment: 
 conda_file: model/environment/conda.yml
 image: mcr.microsoft.com/azureml/openmpi3.1.2-ubuntu18.04:20210727.v1
instance_type: Standard_DS2_v2
instance_count: 1
</code></pre>
Deployment is triggered using the following command:
<pre><code>az deployment create --name geen-depl-2
</code></pre>

Test your understanding of deployments with this quiz. 

Introduction

Getting Started

Creating Azure Resources

Building Azure ML Pipeline

Building a diabetes model pipeline

Service Deployment

Creating a Deep Learning Pipeline

Detect Posture Project

Conclusion

Appendix

Sentiment Analysis Using Natural Language Processing

Quiz on Deployment