-
Book Overview & Buying
-
Table Of Contents
The Kaggle Book - Second Edition
By :
Kaggle notebooks are free to use, but not without limits (more on that in the next chapter, Working and Learning with Notebooks), and the first one you are likely to hit is the time limit – 12 hours for a CPU/GPU session, 9 for TPU. A popular alternative is Google Colab, a free Jupyter Notebook environment that runs entirely in the cloud: https://colab.research.google.com.Even once we’ve moved the computations there, we might still want to have access to the Kaggle datasets, so importing them into Colab is a rather handy feature. The remainder of this section discusses the steps necessary to use Kaggle datasets through Colab.The first thing we do, assuming we are already registered on Kaggle, is head to the account page to generate the API token (an access token containing security credentials for a login session, user identification, privileges, and so on):