Learn Amazon SageMaker by Julien Simon

Author:Julien Simon
Language: eng
Format: epub
Publisher: Packt Publishing Ltd.
Published: 2020-08-26T00:00:00+00:00

Adding libraries for deployment

If you need specific libraries to be available at prediction time, the only option is to customize the framework container. You can pass its name to the estimator with the image_uri parameter:

sk = SKLearn(entry_point='myscript.py', image_uri= '123456789012.dkr.ecr.eu-west-1.amazonaws.com/my-sklearn' . . .

We covered a lot of technical topics in this section. Now, let's look at the big picture.

Putting it all together

The typical workflow when working with framework code looks like this:

Implement Script Mode in your code; that is, the necessary hyperparameters, input data, and output location.

If required, add a model_fn() function to load the model.

Test your training code locally, outside of any SageMaker container.

Configure the appropriate estimator (XGBoost, TensorFlow, and so on).

Train in Local Mode using the estimator, with either the built-in container or a container you've customized.

Deploy in Local Mode and test your model.

Switch to a managed instance type (say, ml.m5.large) for training and deployment.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.