Run TensorFlow Tutorials in the Cloud
How to run TensorFlow tutorials in the cloud using Google Colab for free.
Login to your Google account and click on the links below to run the tutorials:
Tutorial #01 Simple Linear Model:
https://colab.research.google.com/git...
Tutorial #02 Convolutional Neural Network:
https://colab.research.google.com/git...
Tutorial #03-C Keras API:
https://colab.research.google.com/git...
The full list of tutorials is available on GitHub:
https://github.com/Hvass-Labs/TensorF...
You can run the other tutorials in Google Colab by replacing the Notebook filename in the links above.
Some of the tutorials use special Python-modules (source-code files) from the original repository on GitHub, so you first have to clone the repository to your workspace on Google Colab by running these commands in the beginning of the Notebook:
import os
work_dir = "/content/TensorFlow-Tutorials/"
if os.getcwd() != work_dir:
!git clone https://github.com/Hvass-Labs/TensorF...
os.chdir(work_dir)
Some of the tutorials use special Python-packages (for example PrettyTensor) that have to be installed by running the following command in the beginning of the Notebook:
!pip install PrettyTensor
Some of the TensorFlow tutorials require very large downloads (e.g. Tutorial #22 automatically downloads 19 GB of data!). This data will be deleted when you log out of your Google account, so you have to redownload it every time you want to run the Notebook.
Google Colab can be quite unstable especially in GPU-mode, so it is only intended for trying out TensorFlow. Serious work should be done on your own computer or with a more reliable cloud-computing service.