thank you for sharing, i just try azure ml and the look of azure ml is different and i can not find experiment tab. I want to export the predicted value and score model to run in powerBI desktop. I have tried to find out a couple of days but still not suceed. Could you please guide me on how to do this. That would be highly appreciated
@MattEland2 жыл бұрын
It's now called the "Jobs" tab after a recent update! Microsoft will change things from time to time!
@JohnLee-wv4wq8 ай бұрын
can you use this to assist a solution to ask patient set of symptoms and reach a possible disease?
@MattEland8 ай бұрын
That sounds a bit like an expert system or potentially a high risk use of generative AI.
@kevindibb6534 Жыл бұрын
Why is the Automated ML soooo much slower than running a Notebook? Even a simple Automated ML is taking 45 min to 1 1/2 hours, when similar in a Notebook will run in seconds? Am I doing something wrong or is that just how it is?
@oliverharvey75612 жыл бұрын
you called the end-point cost “pretty reasonable” …i was doing my best to set up an automated ml price prediction service for my company and it seemed to be about £100/month on the lowest performance server. that is not attractive to me. it seems possible (though awkward) to run the container on an existing App Service, which hopefully offers a more feasible approach-
@MattEland2 жыл бұрын
It depends on what you need. If you need real-time prediction and AKS, it's going to be more expensive. If you need batched prediction on an ACI, it's going to be less expensive. They're also working on serverless / function versions, it appears, which would be cheaper as well. Of course, you can always use AMLS to train your models and then download them and deploy them outside of these options as well. You could export a model to ONNX, for example, and then load it up in a free-tier app service or even just run it locally.