terraform-provider-azurerm VS cortex

So this is doing my head in. Related to https://github.com/hashicorp/terraform-provider-azurerm/issues/6117

https://registry.terraform.io/providers/hashicorp/azurerm/latest/docs

There are different Terraform Providers that enable Terraform to interact with Microsoft Azure. The most common one are Azure Stack, AzureDevops, AzureRM, AzAPI and AzureAD.. In this tutorial, we use the AzureRM Terraform Provider. Let's create a Terraform file for the AzureRM Terraform Provider.

Yeah we imported all the related resources. I could now find an issue, which exactly describes our problem. Unfortunately it is open since 2020: https://github.com/hashicorp/terraform-provider-azurerm/issues/8794

Like other people said, use the azurerm provider docs, they're pretty good. But that's where knowing Azure comes in handy because you'll have to figure out what TF resource to use to accomplish a given goal.

It’s not difficult at all syntactically. But you must understand the provider you are automating. So your azure knowledge is key in this case. Read the Azure provider docs and you will be easily able to put something together. https://registry.terraform.io/providers/hashicorp/azurerm/latest/docs

I found GPT-2 with PyTorch, Cortex, and Transformers.

Technologies: Flask/FastAPI, Kubernetes, Docker, Cortex, Seldon

oh I see - that's neat. I think cortex.dev had some support for that sort of thing. It is? was? a wrapper to manage an EKS cluster for you and it could do things like hot swap what's in ram, autoscale, or use spot instances to save costs.

I use cortex.dev for deploying my models. I have successfully deployed finetuned T5s, which seems similar to your usecase.

Arghs, my bad, yeah, doesn't seem to be available yet. But maybe Cortex might help: https://github.com/cortexlabs/cortex