Kernel dies during trainer.train() and unable to complete the lab exercise


Figured it out :slight_smile:

Can you report how you fixed it? Because this topic comes up a lot.

Sure, the issue choosing the right instance.

1 Like

I am unable choose any instance other than the default ml.t3.medium. Is there a trick I’m missing?

1 Like

After launching the second lab the default instance type is: ml.t3.medium. I believe the suggested type is: ml.m5.large. However when i select the ml.m5.large instance i get a pop up error rooted in an access denial error based on user role. Why is amazon provision an instance not capable of running the lab? I think this is disappointing from Amazon.

Note: I selected the wrong instance. The notebook says to use: ml.m5.2xlarge which works if you do what it says. Apologies Amazon, I “should have gone to Specsavers”.

Hi everyone. Please check the screenshot in the notebook carefully. I believe the instance type should be ml.m5.2xlarge. Notice the 2x. Please check point 10 of our FAQ to know how to check if you have the correct instance. Hope this helps.

Thanks Chris :+1: