Open rashodqaim opened 3 weeks ago
Switching out GPT models is fairly easy to do on a running IA system. It involves the following steps:
AZURE_OPENAI_CHATGPT_DEPLOYMENT
(required)local.env
AZURE_OPENAI_CHATGPT_MODEL_NAME
AZURE_OPENAI_CHATGPT_MODEL_VERSION
Once the App Service has restarted the web app should be up and using your new model deployment.
NOTE: You will need to manually deploy the "gpt-4o" model to your instance of Azure OpenAI. IA does not currently support deploying gpt-4o model directly.
I think I have follow-up question for this ticket. Yesterday I tried to change the open AI service that I use in the infoasst-web app service. I changed the env variables:
However after the restart I'm getting application error, also on the other hand I noticed that my Azure AI service and previous (created by initial deployment) Azure Open AI service is using the same Endpoint URL.
My question is why am I getting the error after updating the env variables
I am also seeing errors and inconsistencies when updating env variables.
If your Azure OpenAI instance is in another resource group, you may need to assign permissions to read properties from it.
Assign the role "Cognitive Services OpenAI User" to the service principal named "infoasst_mgmt_access_xxxx" that matches your Info Assistant deployment on the resource group where AOAI is or directly on the AOAI instance.
My OpenAI instance is in the same resource group but it's another instance (not the initial one created by the make deploy script)
I have a simple question, I am trying to upgrade the IA tool from gpt-35-turbo-16k to gpt-4o, without creating a whole new IA tool. There is already a lot of data inside the tool and people using it, so I don't want to deploy a new IA tool add the information back in and give everyone access to it again. The CodeSpace I use to create this is shut down as well. Documentation on how to this upgrade would be perfect.