.. _sec_sagemaker:

Using Amazon SageMaker
======================


Many deep learning applications require a significant amount of
computation. Your local machine might be too slow to solve these
problems in a reasonable amount of time. Cloud computing services give
you access to more powerful computers to run the GPU-intensive portions
of this book. This tutorial will guide you through Amazon SageMaker: a
service that allows you to run this book easily.

Registering and Logging In
--------------------------

First, we need to register an account at https://aws.amazon.com/. We
encourage you to use two-factor authentication for additional security.
It is also a good idea to set up detailed billing and spending alerts to
avoid any unexpected surprises in case you forget to stop any running
instance. Note that you will need a credit card. After logging into your
AWS account, go to your `console <http://console.aws.amazon.com/>`__ and
search for "SageMaker" (see :numref:`fig_sagemaker`) then click to
open the SageMaker panel.

.. _fig_sagemaker:

.. figure:: ../img/sagemaker.png
   :width: 300px

   Open the SageMaker panel.



Creating a SageMaker Instance
-----------------------------

Next, let us create a notebook instance as described in
:numref:`fig_sagemaker-create`.

.. _fig_sagemaker-create:

.. figure:: ../img/sagemaker-create.png
   :width: 400px

   Create a SageMaker instance.



SageMaker provides multiple `instance
types <https://aws.amazon.com/sagemaker/pricing/instance-types/>`__ of
different computational power and prices. When creating an instance, we
can specify the instance name and choose its type. In
:numref:`fig_sagemaker-create-2`, we choose ``ml.p3.2xlarge``. With
one Tesla V100 GPU and an 8-core CPU, this instance is powerful enough
for most chapters.

.. _fig_sagemaker-create-2:

.. figure:: ../img/sagemaker-create-2.png
   :width: 400px

   Choose the instance type.





.. raw:: html

     <div class="mdl-tabs mdl-js-tabs mdl-js-ripple-effect"><div class="mdl-tabs__tab-bar text"><a href="#mxnet-1-0" onclick="tagClick('mxnet'); return false;" class="mdl-tabs__tab is-active">mxnet</a><a href="#pytorch-1-1" onclick="tagClick('pytorch'); return false;" class="mdl-tabs__tab ">pytorch</a><a href="#tensorflow-1-2" onclick="tagClick('tensorflow'); return false;" class="mdl-tabs__tab ">tensorflow</a></div>



.. raw:: html

     <div class="mdl-tabs__panel is-active" id="mxnet-1-0">

A Jupyter notebook version of this book for fitting SageMaker is
available at https://github.com/d2l-ai/d2l-en-sagemaker. We can specify
this GitHub repository URL to let SageMaker clone this repository during
instance creation, as shown in :numref:`fig_sagemaker-create-3`.



.. raw:: html

     </div>



.. raw:: html

     <div class="mdl-tabs__panel " id="pytorch-1-1">

A Jupyter notebook version of this book for fitting SageMaker is
available at https://github.com/d2l-ai/d2l-pytorch-sagemaker. We can
specify this GitHub repository URL to let SageMaker clone this
repository during instance creation, as shown in
:numref:`fig_sagemaker-create-3`.



.. raw:: html

     </div>



.. raw:: html

     <div class="mdl-tabs__panel " id="tensorflow-1-2">

A Jupyter notebook version of this book for fitting SageMaker is
available at https://github.com/d2l-ai/d2l-tensorflow-sagemaker. We can
specify this GitHub repository URL to let SageMaker clone this
repository during instance creation, as shown in
:numref:`fig_sagemaker-create-3`.



.. raw:: html

     </div>



.. raw:: html

     </div>

.. _fig_sagemaker-create-3:

.. figure:: ../img/sagemaker-create-3.png
   :width: 400px

   Specify the GitHub repository.



Running and Stopping an Instance
--------------------------------

It may take a few minutes before the instance is ready. When it is
ready, you can click on the "Open Jupyter" link as shown in
:numref:`fig_sagemaker-open`.

.. _fig_sagemaker-open:

.. figure:: ../img/sagemaker-open.png
   :width: 400px

   Open Jupyter on the created SageMaker instance.



Then, as shown in :numref:`fig_sagemaker-jupyter`, you may navigate
through the Jupyter server running on this instance.

.. _fig_sagemaker-jupyter:

.. figure:: ../img/sagemaker-jupyter.png
   :width: 400px

   The Jupyter server running on the SageMaker instance.



Running and editing Jupyter notebooks on the SageMaker instance is
similar to what we have discussed in :numref:`sec_jupyter`. After
finishing your work, do not forget to stop the instance to avoid further
charging, as shown in :numref:`fig_sagemaker-stop`.

.. _fig_sagemaker-stop:

.. figure:: ../img/sagemaker-stop.png
   :width: 300px

   Stop a SageMaker instance.



Updating Notebooks
------------------



.. raw:: html

     <div class="mdl-tabs mdl-js-tabs mdl-js-ripple-effect"><div class="mdl-tabs__tab-bar text"><a href="#mxnet-3-0" onclick="tagClick('mxnet'); return false;" class="mdl-tabs__tab is-active">mxnet</a><a href="#pytorch-3-1" onclick="tagClick('pytorch'); return false;" class="mdl-tabs__tab ">pytorch</a><a href="#tensorflow-3-2" onclick="tagClick('tensorflow'); return false;" class="mdl-tabs__tab ">tensorflow</a></div>



.. raw:: html

     <div class="mdl-tabs__panel is-active" id="mxnet-3-0">

We will regularly update the notebooks in the
`d2l-ai/d2l-en-sagemaker <https://github.com/d2l-ai/d2l-en-sagemaker>`__
GitHub repository. You can simply use the ``git pull`` command to update
to the latest version.



.. raw:: html

     </div>



.. raw:: html

     <div class="mdl-tabs__panel " id="pytorch-3-1">

We will regularly update the notebooks in the
`d2l-ai/d2l-pytorch-sagemaker <https://github.com/d2l-ai/d2l-pytorch-sagemaker>`__
GitHub repository. You can simply use the ``git pull`` command to update
to the latest version.



.. raw:: html

     </div>



.. raw:: html

     <div class="mdl-tabs__panel " id="tensorflow-3-2">

We will regularly update the notebooks in the
`d2l-ai/d2l-tensorflow-sagemaker <https://github.com/d2l-ai/d2l-tensorflow-sagemaker>`__
GitHub repository. You can simply use the ``git pull`` command to update
to the latest version.



.. raw:: html

     </div>



.. raw:: html

     </div>

First, you need to open a terminal as shown in
:numref:`fig_sagemaker-terminal`.

.. _fig_sagemaker-terminal:

.. figure:: ../img/sagemaker-terminal.png
   :width: 300px

   Open a terminal on the SageMaker instance.



You may want to commit your local changes before pulling the updates.
Alternatively, you can simply ignore all your local changes with the
following commands in the terminal.



.. raw:: html

     <div class="mdl-tabs mdl-js-tabs mdl-js-ripple-effect"><div class="mdl-tabs__tab-bar text"><a href="#mxnet-5-0" onclick="tagClick('mxnet'); return false;" class="mdl-tabs__tab is-active">mxnet</a><a href="#pytorch-5-1" onclick="tagClick('pytorch'); return false;" class="mdl-tabs__tab ">pytorch</a><a href="#tensorflow-5-2" onclick="tagClick('tensorflow'); return false;" class="mdl-tabs__tab ">tensorflow</a></div>



.. raw:: html

     <div class="mdl-tabs__panel is-active" id="mxnet-5-0">

.. raw:: latex

   \diilbookstyleinputcell

.. code:: bash

    cd SageMaker/d2l-en-sagemaker/
    git reset --hard
    git pull



.. raw:: html

     </div>



.. raw:: html

     <div class="mdl-tabs__panel " id="pytorch-5-1">

.. raw:: latex

   \diilbookstyleinputcell

.. code:: bash

    cd SageMaker/d2l-pytorch-sagemaker/
    git reset --hard
    git pull



.. raw:: html

     </div>



.. raw:: html

     <div class="mdl-tabs__panel " id="tensorflow-5-2">

.. raw:: latex

   \diilbookstyleinputcell

.. code:: bash

    cd SageMaker/d2l-tensorflow-sagemaker/
    git reset --hard
    git pull



.. raw:: html

     </div>



.. raw:: html

     </div>

Summary
-------

-  We can launch and stop a Jupyter server through Amazon SageMaker to
   run this book.
-  We can update notebooks via the terminal on the Amazon SageMaker
   instance.

Exercises
---------

1. Try to edit and run the code in this book using Amazon SageMaker.
2. Access the source code directory via the terminal.

`Discussions <https://discuss.d2l.ai/t/422>`__