Search code examples
amazon-web-servicesdockerartificial-intelligenceamazon-sagemakerlarge-language-model

How can I determine the Docker image used by a SageMaker Endpoint created with JumpStart


I have created an LLM Endpoint using the SageMaker Jumpstart Service. I've tested it out, and everything works great. But where should I check which Docker image is used?

For instance, in the Endpoint Configuration, I can only find these variables:

Key                                                     Value


sagemaker:user-profile-arn                               My User id
    
aws-jumpstart-inference-model-uri                        s3://jumpstart-cache-prod-us-east-1/meta-infer/infer-meta-textgeneration-llama-2-7b.tar.gz

sm-jumpstart-monitor                                     True

sm-jumpstart-id                                          JumpStart ID

sagemaker:domain-arn                                     domain

I do need the docker image to deploy Endpoint via CloudFormation.

Regards, Volodymyr


Solution

  • Amazon SageMaker JumpStart is just a quick way to start a regular endpoint.

    Got to the relevant Model - In the console, select SageMaker service, then in the left tab, open Inference and Model.

    In the model configuration, you should see Image URI.