Skip to content
By using AWS re:Post, you agree to the AWS re:Post Terms of Use
AWS re:Postre:Post
About re:Post

Can I mount EFS to a custom Sagemaker inference endpoint?

0

I've deployed an ML model using a Sagemaker inference endpoint using a custom Docker image.

Currently, the model artifacts are packaged as a .tar.gz file and pulled from S3. However, I'd like to pull the model artifacts from EFS. Can I mount EFS to my Docker container with Sagemaker?

I'm aware that there are ways to mount EFS to a Sagemaker notebook, but can't find anything about inference endpoints.

Topics
Machine Learning & AIStorageContainers
Tags
Amazon SageMakerAmazon Elastic File SystemMachine Learning & AIContainers
Language
English
asked 2 years ago738 views
2 Answers
0

Hi,

This articles describes how to mount an EFS drive onto a Sagemaker docker image via the Sagemaker APIs: https://www.cnblogs.com/jesse123/p/16300765.html

Best

Didier

EXPERT
answered 2 years ago
  • lkbill
    2 years ago

    I had a similar requirement to use EFS mount along with sagemaker hosted inference endpoint than s3.I didn't find a supported solution. if I am not wrong, in the shared url, sagemaker API works only with training job and not necessarily specially if it's real time inference hosted via sagemaker endpoint.

  • Julia
    2 years ago

    Thanks, but as @lkbill mentioned, the solution you reference is for training. I'm looking for an inference solution.

0

Hi,

So far, EFS is not supported for processing jobs and inference job. Howerver, if your use case does not require GPUs and can do with less than 10GB of memory per instance, maybe you can perform inference using Lambda containers and layers. Reference: https://aws.amazon.com/cn/blogs/machine-learning/build-reusable-serverless-inference-functions-for-your-amazon-sagemaker-models-using-aws-lambda-layers-and-containers/

AWS
answered 2 years ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Relevant content