"Kernel has died"

Course Q&A Generative AI with Large Language Models GenAI with LLMs Resources
1

In the labs while trying to train the models, I am getting error “the kernel has died” frequently and not able to proceed due to the same. Is anyone else getting the same error? Any suggestion on what to do?

2

Yes try filling up this form:

3

Please check if you selected the right size for your virtual machine. The recommended instance is Ml.m5.2xlarge, and it will create a 8 CPU 32 GB VM.

4

I had the same problem with Lab 2.

I was unable to select any kernel instance with > 4GB RAM, so the train() functions would fail after exausting the 4GB RAM.

Failed to start kernelFailed to launch app [sagemaker-data-scienc-ml-t3-xlarge-926c947fdea9528b6cf58021c71b]. AccessDeniedException: User: arn:aws:sts::452237133870:assumed-role/sagemaker-studio-vpc-firewall-us-east-1-sagemaker-execution-role/SageMaker is not authorized to perform: sagemaker:CreateApp on resource: arn:aws:sagemaker:us-east-1:452237133870:app/d-8p6dilptvyvi/sagemaker-user-profile-us-east-1/kernelgateway/sagemaker-data-scienc-ml-t3-xlarge-926c947fdea9528b6cf58021c71b with an explicit deny in an identity-based policy (Context: RequestId: cc67018b-a996-462a-a88f-e20588b28da6, TimeStamp: 1697733015.3176513, Date: Thu Oct 19 16:30:15 2023)

5

You can only select the bigger instance with the exact name shown in the lab instructions. When I did it, the name was: ml.m5.2xlarge.

6

Thanks, Leonardo. Must have misread that!