Google Colab Guide

by Prof. Seungchul Lee
Industrial AI Lab at POSTECH

Table of Contents

1. Overview

In [1]:
<center><iframe src="" 
width="560" height="315" frameborder="0" allowfullscreen></iframe></center>
In [2]:
<center><iframe src="" 
width="560" height="315" frameborder="0" allowfullscreen></iframe></center>

2. Requirement

  • You might need to have google account for Colab

  • Owing to its remote access, there are no particular computing specifications.

  • Saving the work regularly prevents the kernel from disconnecting on long-time work

3. Connect and Create New Notebook

  • You'll be able to create ipynb file by clicking NEW PYTHON 3 NOTEBOOK in the right below at main page
    • Every ipynb file would be saved in GoogleDrvie even if it's not synchronized to it.

4. GPU Setting

  • Tesla K80 12GB available for free
  • Available 12 hours in a row and automatically disconnected if there is no response 90 mintues later
    • Not meaning the 12-hour regulation in a day but meaning the 12-hour sequential limitation
  • Select Runtime -> Change runtime type in the left above

  • Click Hardware accelerate, select GPU and save it

5. Mounting Your Google Drive

  • Synchronization makes it easier to use the data on GoogleDrive

  • Type the code as above and implement it

In [ ]:
from google.colab import auth

from google.colab import drive