Want to infrequently train an ML model on demand - what service do I use?

Jim Gale 1 Reputation point Microsoft Employee
2021-02-03T18:26:02.803+00:00

Training an ML model takes ~50 mins + 6 gb once a week. I have specific C# code it needs - can't use python/R.

Can't do Azure Functions since it is a 10min/1.5gig limit. Don't want to provision an Azure App Service b/c I only use it once a week - would be expensive for that amount of GB. Can't do a VM b/c it would be a security risk even IF I wanted to set up a start/stop to not get charged, which I don't.

Can a pipeline/build-service handle 50 mins & 6 gb? Or, what would be a better method to run what /would/ be an Azure Function for ~50 minutes, using 6gb on demand?

Azure Functions
Azure Functions
An Azure service that provides an event-driven serverless compute platform.
4,307 questions
.NET Machine learning
.NET Machine learning
.NET: Microsoft Technologies based on the .NET software framework.Machine learning: A type of artificial intelligence focused on enabling computers to use observed data to evolve new behaviors that have not been explicitly programmed.
150 questions
0 comments

1 answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. Pramod Valavala 20,591 Reputation points Microsoft Employee
    2021-02-04T06:39:29.493+00:00

    Azure Container Instances is one option that you could consider. You could have a simple Logic App to run the job (using the ACI connector) based on the required schedule or just on HTTP request.

    There are GPU Containers available too for you to leverage.