|
| 1 | +# Post Deployment Configuration |
| 2 | + |
| 3 | +## Step 1: Execute the Script |
| 4 | + |
| 5 | +### 1.1 Open PowerShell, change directory where you code cloned, then run the deploy script: |
| 6 | + |
| 7 | +``` |
| 8 | +cd .\Deployment\ |
| 9 | +``` |
| 10 | + |
| 11 | +### 1.2 Choose the appropriate command based on your deployment method: |
| 12 | + |
| 13 | +**If you deployed using `azd up` command:** |
| 14 | +``` |
| 15 | +.\resourcedeployment.ps1 |
| 16 | +``` |
| 17 | + |
| 18 | +**If you deployed using custom templates, ARM/Bicep deployments, or `az deployment group` commands:** |
| 19 | +``` |
| 20 | +.\resourcedeployment.ps1 -ResourceGroupName "<your-resource-group-name>" |
| 21 | +``` |
| 22 | + |
| 23 | +> **Note:** Replace `<your-resource-group-name>` with the actual name of the resource group containing your deployed Azure resources. |
| 24 | +
|
| 25 | +> **💡 Tip**: Since this guide is for azd deployment, you'll typically use the first command without the `-ResourceGroupName` parameter. |
| 26 | +
|
| 27 | +If you run into issue with PowerShell script file not being digitally signed, you can execute below command: |
| 28 | + |
| 29 | +``` |
| 30 | +powershell.exe -ExecutionPolicy Bypass -File ".\resourcedeployment.ps1" |
| 31 | +``` |
| 32 | + |
| 33 | +### 1.3 You will be prompted for the following parameters with this Screen : |
| 34 | + |
| 35 | +<img src="./images/deployment/Deployment_Input_Param_01.png" width="900" alt-text="Input Parameters"> |
| 36 | + |
| 37 | +#### 1.3.1 **Email** - used for issuing certificates in Kubernetes clusters from the [Let's Encrypt](https://letsencrypt.org/) service. Email address should be valid. |
| 38 | + |
| 39 | +<img src="./images/deployment/Deployment_Login_02.png" width="900" alt-text="Login"> |
| 40 | + |
| 41 | +#### 1.3.2 You will be prompted to Login, Select a account and proceed to Login. |
| 42 | + |
| 43 | +#### 1.3.3 **GO !** - Post Deployment Script executes Azure Infrastructure configuration, Application code compile and publish into Kubernetes Cluster. |
| 44 | + |
| 45 | +#### 1.3.4 Deployment Complete |
| 46 | +#### 🥳🎉 First, congrats on finishing Deployment! |
| 47 | +Let's check the message and configure your model's TPM rate higher to get better performance. |
| 48 | +You can check the Application URL from the final console message. |
| 49 | +Don't miss this Url information. This is the application's endpoint URL and it should be used for your data importing process. |
| 50 | + |
| 51 | +<img src="./images/deployment/Deployment_Screen02.png" alt="Success Deployment" width="900"> |
| 52 | + |
| 53 | +### Manual Deployment Steps: |
| 54 | +**Create Content Filter** - Please follow below steps |
| 55 | +> * Navigate to project in Azure OpenAI, then go to Azure AI Foundry, select Safety + security |
| 56 | +> * Click on Create Content Filter and set the filters to a high threshold for the following categories: |
| 57 | + ``` |
| 58 | + Hate, Sexual, Self-harm, Violence |
| 59 | + ``` |
| 60 | +> * Please select the checkbox of profanity |
| 61 | +> * Leave all other configurations at their default settings and click on create |
| 62 | +
|
| 63 | +## Step 2: Configure Azure OpenAI Rate Limits |
| 64 | + |
| 65 | +> **Capacity Note:** |
| 66 | +> * The deployment script creates models with a setting of 1 token per minute (TPM) rate limit. |
| 67 | +> * Faster performance can be achieved by increasing the TPM limit with Azure AI Foundry. |
| 68 | +> * Capacity varies for [regional quota limits](https://learn.microsoft.com/en-us/azure/ai-services/openai/quotas-limits#regional-quota-limits) as well as for [provisioned throughput](https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/provisioned-throughput). |
| 69 | +> * As a starting point, we recommend the following quota threshold be set up for this service run. |
| 70 | +
|
| 71 | +| Model Name | TPM Threshold | |
| 72 | +|------------------------|---------------| |
| 73 | +| GPT-4.1-mini | 100K TPM | |
| 74 | +| text-embedding-3-large | 200K TPM | |
| 75 | + |
| 76 | + |
| 77 | +> **⚠️ Warning:** **Insufficient quota can cause failures during the upload process.** Please ensure you have the recommended capacity or request for additional capacity before start uploading the files. |
| 78 | +
|
| 79 | + |
| 80 | +### 2.1. Browse to the project in Azure AI Foundry, and select **each of the 2 models** within the `Deployments` menu: |
| 81 | +<img src="./images/deployment/Control_Model_TPM000.png" alt="Select Model" width="700"> |
| 82 | + |
| 83 | +### 2.2. Increase the TPM value for **each model** for faster report generation: |
| 84 | +<img src="./images/deployment/Control_Model_TPM001.png" alt="Set Token per minute" width="700"> |
| 85 | + |
| 86 | +### 3. Data Uploading and Processing |
| 87 | +After increasing the TPM limit for each model, let's upload and process the sample documents. |
| 88 | + |
| 89 | +Execute this command: |
| 90 | + |
| 91 | +<img src="./images/deployment/Deployment_last_step.png" alt="Set Token per minute" width="700"> |
0 commit comments