Why use Google Colab?

You can use the Jupyter notebook on your local computer. Goggle Jupyter improves on the Jupyter notebook in many ways. I list the seven most powerful reasons to use Google Colab:

  1. You can get any public Jupyter notebook from a GitHub repository.
  2. You load, edit and save any .ipynbfile to the Google Drive associated with the Colab login. It is helpful to have a separate Google account for each project and thus a different Google Drive.

Note: You can create a git account for any project folder on Google Drive. Each team member hosts on a variety of different local computers. All they need is a browser and internet connection. Your team can be fully distributed gographically by the cloud.

3. You can provision one of many generations of the NVIDIA GPU.

4. You can provision one of many generations of the Google TPU.

5. You can provision a multi-core CPU.

6. Google Colab is free; also, you can upgrade to a premium version of $9.99 per month per account.

7. Colab notebook has many useful extensions of Jupyter notebook.

Loading a file into Colab from a GitHub repo.

In a browser, you can put https://colab.research.google.com, or click here, and a new tab pointing to Google Colab opens:

Image for post
Figure 1: Entrance to Google Colab.

If you are new to Colab, the only file in Recent is Welcome to Collaboratory . As I am not new, you count five files in Recent.

Loading Files From GitHub

Click on GitHub and observe:

Image for post
Figure 2: Clicking on GitHub.

You enter the GitHub account you want to browse. I entered bcottman, my GitHub top-level repository.

Image for post
Figure 3:Choosing bcottman/paso repo

From the repo bcottman/pasoI chose the file bcottman/paso/integration_test_pre_all.ipynb .

Image for post
Figure 4: Choosing the file bcottman/paso/integration_test_pre_all.ipynb.

Resulting in file bcottman/paso/integration_test_pre_all.ipynb being loaded into GCP (Google Cloud Platform) Colab browser.

Image for post
Figure 5. File bcottman/paso/integration_test_pre_all.ipynb loaded into Google Colab.

Loading Files From Google Drive

You can create Google Drive and then create a git account for any project folder. You are going to https://colab.research.google.com. Mounts the Google Drive of the Google account. All files in Google Drive appear when you click on File|Open notebook:

Image for post
Figure 6: Click on File|Open notebook.

Selected the colab_itils.ipynb file:

Image for post
Figure 7. Top-level files in Google Drive.

Provision one of many generations of the NVIDIA GPU.

Click on Runtime|Change runtime type: to provision an Nvidia GPU:

Image for post
Figure 8: Click on Runtime|Change runtime type: to provision an Nvidia GPU
Image for post
Figure 9: Select GPU.

The status of provisioning of an Nvidia GPU is checked with:

The from tensorflow.python.client import device_lib
device_lib.list_local_devices()

output=>

[name: "/device:CPU:0"  
device_type: "CPU"
memory_limit: 268435456
locality { }
incarnation: 17311008600223054265,
name: "/device:GPU:0"
device_type: "GPU"
memory_limit: 14674281152
locality {
bus_id: 1
links { } }
incarnation: 7680686309919727928
physical_device_desc: "device: 0,
name: Tesla T4, pci bus id: 0000:00:04.0, compute capability: 7.5"]

An Nvidia Telsa T4 with 14.67 GB of fast memory is provided.

Provision Google TPU.

Click on: to provision a Google TPU. Select TPU.

Image for post
Figure 10: Select TPU.

Conclusion

In this short blog article, I showed:

  1. You can get any public file from a GitHub repository.
  2. You load, edit and save any .ipynbfile to the Google Drive associated with the Colab login.

3. You can provision one of many generations of the NVIDIA GPU.

4. You can provision one of many generations of the Google TPU.

One critical capability with Google Colab is that team members can collaborate on a project using shared files on GitHub. Also, each team member can create their development sandbox on their own Google Drive.

The cloud empowers remote work. Happy coding!