Skip to Content

Setup AWS - Create a SageMaker Notebook

test
0 %
Setup AWS - Create a SageMaker Notebook
Details
// Explore More Tutorials

Setup AWS - Create a SageMaker Notebook

09/03/2019

Create a SageMaker Notebook instance that will be used to complete this tutorial series.

You will learn

  • Access Amazon SageMaker Management Console
  • Create a SageMaker Notebook instance

During this tutorial series, you will be using a SageMaker Notebook as your primary tool to not only leverage the SageMaker platform to train your TensorFlow model.

Notebooks are powerful tools that can be used for multiple purposes.

For example in the current scenario, you will use Notebooks to analyze your dataset using Python code, but also to interact with S3 buckets, trigger a SageMaker training job or deploy your trained model as part of a using a TensorFlow Serving Docker image.

To find out more about Amazon SageMaker, please check the following URL: https://aws.amazon.com/sagemaker/

Step 1: Access the SageMaker Console

Access the Amazon SageMaker Console (you also use the search for SageMaker in the Amazon Web Services Management Console).

Amazon Web Services
Log on to answer question
Step 2: Create a Notebook instance

First, you will need to create a SageMaker Notebook instance.

On the left side, select Notebook instances, then click on Create notebook instance.

Amazon Web Services

Enter the following Notebook instance settings details:

  • Notebook instance name: sap-hxe-eml-demo
Amazon Web Services

Enter the following Permissions and encryption details:

  • IAM role: Enter a custom IAM role ARN
  • Custom IAM role ARN: enter the ARN for the sap-hxe-eml-sagemaker-role role create previously
Amazon Web Services

Scroll down, then click on Create notebook instance.

The process will take up to a few minutes to complete.

Amazon Web Services

Once In Service, click on Open JupyterLab.

Amazon Web Services

Provide an answer to the question below then click on Validate.

Paste the notebook URL.
×
Step 3: Check the Python and TensorFlow in a Console

Now that you have accessed your JupyterLab notebook, let’s check the Python and TensorFlow version in a IPython console.

Amazon Web Services

On the menu bar, select File > New > Console.

Amazon Web Services

Select conda_tensorflow_p36 as Kernel then click on Select.

Amazon Web Services

This will open an interaction Python environment.

Once open, paste the following code in the input field:

import sys, tensorflow as tf
print("Python version: {}".format(sys.version))
print("TensorFlow version: {}".format(tf.__version__))
Amazon Web Services

Then press SHIFT + ENTER to execute the command.

Amazon Web Services

Provide an answer to the question below then click on Validate.

Paste the command output.
×
Step 4: Check the Python and TensorFlow in a Terminal

Let’s now do the same in a Terminal console.

On the menu bar, select File > New > Terminal.

Amazon Web Services

This will open a terminal console with bash.

Once open, paste the following series of commands:

source activate tensorflow_p36
python -c 'import tensorflow as tf; print(tf.__version__)'
python --version

Note: To paste the content of your current clipboard, press SHIFT + INSERT instead of CTRL + V and use CTRL + INSERT instead of CTRL + C.

Amazon Web Services
Paste the command last output. Refresh page if validation times out.
×

Next Steps

Back to top