Use of OpenAI GPT models in translate5
Azure enables direct access to OpenAI models in the Microsoft Azure Cloud via the Azure OpenAI Service. On this page you will find information on configuring and setting up the plug-in for this setup.
Settings in the system configuration
Translate5’s Azure integration has its own specific configuration based on Azure’s requirements to work with what translate5 offers for GPT. After activating the OpenAI plug-in, you need to go to the translate5 system configuration under “System setup: Language resources”, and configure the settings as listed below.
| All configuration values are also available on https://portal.azure.com. If you already have an app with a model in use, you can use the existing configuration values. If not, you can generate the necessary resources via this workflow. |
|---|
| Designation | System configuration ID in translate5 | Explanation | Where can I find the relevant information on the Azure Cloud? |
|---|---|---|---|
| Subscription ID | runtimeOptions.plugins.OpenAI.Azure.subscriptionId | ID of the subscription used for the GPT application in Azure | Home > Subscriptions |
| Resource group name | runtimeOptions.plugins.OpenAI.Azure.resourceGroupName | Name of the resource group used for the GPT application in Azure | Search > Resource groups > Create |
| Account name | runtimeOptions.plugins.OpenAI.Azure.accountName | Account name that is also used as part of the individual domain name in Azure | Navigate to ai.azure.com and select the desired application (if more than one is available). You can now find the account name in the address bar of the browser, after the “accounts/” part. |
Tenant ID | runtimeOptions.plugins.OpenAI.Azure.tenantId | ID of the (tenant) directory | Search > Microsoft Entra ID > Manage > App registrations > %YOUR APP% > Overview |
Client ID | runtimeOptions.plugins.OpenAI.Azure.clientId | ID of the (client) application | Search > Microsoft Entra ID > Manage > App registrations > %YOUR APP% > Overview |
Client secret | runtimeOptions.plugins.OpenAI.Azure.clientSecret | Client Secret for access to resource management | Search > Microsoft Entra ID > Manage > App registrations > %YOUR APP% > Manage > Certificates and secrets |
Deployment type | runtimeOptions.plugins.OpenAI.Azure.deploymentType | Deployment type (Stock Keeping Unit, SKU) | See https://learn.microsoft.com/en-us/azure/ai-foundry/model-inference/concepts/deployment-types; usually “standard”. |
Server | runtimeOptions.plugins.OpenAI.server | API endpoint to which chat completion requests are sent | https://ai.azure.com/> Home >Azure OpenAI endpoint (Important: the end of the URL must include “/openai”) |
API token | runtimeOptions.plugins.OpenAI.apiToken | API access token | https://ai.azure.com/> Home >API key (1/2) |
| After creating a language resource, translate5 automatically creates a model deployment of the selected Azure Cloud model instance. The name of the deployment contains the ID of the translate5 language resource to enable easy identification. Even after a GPT language resource is deleted, the underlying model remains available. (Automated deletion could be implemented if this appears useful.) |
|---|---|
| During the training of a model, a new deployment is created once the training process is completed. Please note that the Azure documentation says: “After you deploy a customized model, if at any time the deployment remains inactive for greater than fifteen (15) days, the deployment is deleted. The deployment of a customized model is inactive if the model was deployed more than fifteen (15) days ago and no completions or chat completions calls were made to it during a continuous 15-day period. The deletion of an inactive deployment doesn’t delete or affect the underlying customized model, and the customized model can be redeployed at any time.” |
Workflow to create all necessary resources in Azure
- Create a subscription via “Home” > “Subscriptions” > “Add”
- Create a resource group via “Search” > “Resource groups” > “Create”
- Create a resource via “All resources”> “Create” > “AI + Machine learning” > “Azure OpenAI” > “Create”