Programming Exercises

 

Most self-study courses contain companion programming exercises. These exercises provide practice in applying machine learning concepts to solve real-world problems. In each exercise, you’ll use TensorFlow to experiment with machine learning models in Python.

Prerequisites

No prior TensorFlow knowledge is required to complete the exercises, but some familiarity with Python programming basics is highly recommended. 

What TensorFlow APIs do the programming exercises use?

The programming exercises rely on several high-level TensorFlow APIs, including Estimators and tf.keras. To learn more about Estimators, see the Premade Estimators page on tensorflow.org. To learn more about tf.keras, see the Tutorials section of tensorflow.org.

Will the exercises be updated when new versions of TensorFlow are released?

Yes. Exercises will be regularly updated to ensure they are always compatible with the latest TensorFlow API release.

How do I run the exercises?

You can run the exercises in your browser (no installation required) using the Colaboratory platform. Just click on the programming exercise links on the course pages to launch them in Colaboratory. 

What are the system requirements for Colaboratory?

See the Colaboratory FAQs for a full list of system requirements.

Where can I get a list of the programming exercises?

See the Exercises page for a set of links to all programming exercises.

Can I download the exercises and run them locally?

Yes! All exercises are available for download from Machine Learning Crash Course’s GitHub repository in Jupyter notebook (.ipynb) format. See “Running Datalab Locally” for instructions on how to run the exercises on a local machine using the Datalab platform.

Translations

Translations of the exercises are currently available in French, Spanish, Korean, and Simplified Chinese. Note: in all translations, code excerpts are in English. 

Was this helpful?
How can we improve it?
Search
Clear search
Close search
Google apps
Main menu
Search Help Center
false
false