search

Azure-Samples / azure-search-openai-demo

A sample app for the Retrieval-Augmented Generation pattern running in Azure, using Azure AI Search for retrieval and Azure OpenAI large language models to power ChatGPT-style and Q&A experiences.
https://azure.microsoft.com/products/search
MIT License
5.63k stars 3.77k forks source link

Cannot build using parms for GPT4, 1106-preview for westus region. #1239

Open evogelpohl opened 5 months ago

evogelpohl commented 5 months ago

Please provide us with the following information:

This issue is for a: (mark with an x)

- [ ] bug report -> please search issues before submitting
- [ x] feature request
- [ ] documentation issue or request
- [ ] regression (a behavior that used to work and stopped in a new release)

Minimal steps to reproduce

I want to use GPT-4, not 3.5 for my deployment. Not vision.

I've confirmed I have quota to deploy OpenAI GPT 4 in westus.

I've edited the list of regions in the main.bicep to include westus for param openAiResourceGroupLocation

I've edited the main.bicep file to change the deployment to:

param chatGptModelName string = (openAiHost == 'azure') ? 'gpt-4' : 'gpt-4' param chatGptModelVersion string = '1106-preview'

This is according to the table @ https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models#gpt-4-and-gpt-4-turbo-preview.

Run azd up & select westus for my openai deployment, and I get this error:

`ERROR: deployment failed: failing invoking action 'provision', error deploying infrastructure: deploying to subscription:

Deployment Error Details: InvalidTemplateDeployment: The template deployment 'openai' is not valid according to the validation procedure. The tracking id is 'e2d3af54-fbfc-49c7-9c40-67cda6153479'. See inner errors for details. DeploymentModelNotSupported: The model 'Format: OpenAI, Name: gpt-4, Version: 1106-preview' of account deployment is not supported.

TraceID: 8874dd9589a4b507224477a8611845ed`

azd version 1.5.1 (commit 3856d1e98281683b8d112e222c0a7c7b3e148e96)

pamelafox commented 5 months ago

Are you able to create that deployment from the Azure Portal as well? Sometimes there are availability issues that we haven't accounted for in our Bicep.

evogelpohl commented 5 months ago

Are you able to create that deployment from the Azure Portal as well? Sometimes there are availability issues that we haven't accounted for in our Bicep.

Yes. I can manually create the OpenAI instance (in the portal, for westus). Then, within the studio, deploy the GPT-4, ver 1106-Preview.

Then, I can define the azd env ... vars as documented specifying a pre-established OAI service, rg, and chat model by name. The azd up process fails w/ the error of ~ 'you cannot change models` or similar.

I'm in a 'catch-22' If I deploy to eastus, we have no quota or permission to use any gpt-4 model at all and the permission form doesn't allow you to specific the region,

If I deploy to westus, it doesn't support deploying gpt-4 using a version that this project accepts. Thus, i'm stuck w/ 3.5-turbo.