LAURENCE MORONEY: Hi, and welcome to part two

of this series on using Google Colab to code, train, and test

neural networks in the browser without needing to install

any kind of a runtime.

In the previous video, Jake introduced you to Colab

and how you can get started with it.

We'll take that to the next level in this video,

and I'm going to show you how to install TensorFlow into Colab

and then play with the different versions of TensorFlow,

such as those that use the GPU.

The first thing you can do is to create a new Colab.

Did you know that you can do this directly

from Google Drive?

Just click New, and you will find

Colaboratory in the More menu.

Once the Colab is created, you'll

have a new code cell ready for you.

You can then type Python code into this.

But I'll create a new code cell and, in it, I

will use the PIP command to install TensorFlow.

PIP is a command line tool, not a code command,

so you have to prefix it with an exclamation mark.

That way Colab will understand it.

When I run the install command, the environment

will download and install TensorFlow

if it isn't already installed.

You can then check if it worked by printing out the installed

version.

This is done with this code.

And when I execute this code, you

can see that we have version 1.12 installed.

Now TensorFlow also comes in different versions.

For example, if you want to use a GPU for learning,

there is a GPU version.

Let's look at how to install that in a fresh notebook.

Because the GPU version needs a GPU,

we first need to be sure that our runtime has a GPU.

So to do this, I find it best to reset all runtimes

and then set the runtime type to be a GPU-based one.

It may take a few seconds to allocate

the new virtual machine and then launch it.

Now I can install the GPU flavor of TensorFlow

with pip install tensorflow-GPU.

You'll see that it isn't present on this machine,

so it will download and install it.

And now I can check the version of TensorFlow as before,

and we'll see that we also have version 1.12.

And that's it.

In the next video in this series,

I'll take that to the next level and show you

how to train a neural network using

Keras to perform classification of breast cancer data.

And I'll see you there.

[MUSIC PLAYING]