How To Get Free VPS, RDP Gui Lifetime Google Colaboratory

akuh

How To Get Free VPS, RDP Gui Lifetime Google Colaboratory - Within the world of Data Science, there are very interesting initiatives, and one of the most interesting ones, in addition to all the training options and tools available, is the Google Colab.

Colab is a cloud service, based on Jupyter Notebooks, which allows the free use of Google GPUs and TPUs, with libraries such as Scikit-learn, PyTorch, TensorFlow, Keras, and OpenCV. All with under Python 2.7 and 3.6, which are not yet available for R and Scala.

Although it has some limitations, which can be consulted on its FAQ page, it is an ideal tool, not only to practice and improve our knowledge in Data Science techniques and tools but also for the development of machine learning applications (pilots) and deep learning, without having to invest in hardware or Cloud resources.

With Colab we can create notebooks or import those that we already have created, as well as share and export them whenever we want. This fluidity when handling information is also applicable to the data sources that we use in our projects (notebooks) so that we can work with the information contained in our own Google Drive, local storage unit, GitHub, and even in others. cloud storage systems, such as Amazon's S3.

Starting to work with Colab

In order to have our workspace in Colab, we will have to have a google account and access the Google Drive service. Once inside, we will give New> Folder, putting the name that we consider

Loading the data

When loading the data, which will be used in the notebook, there are several options, but here we will see three:

Enable access to our Google Drive

Upload your data from an Amazon S3 instance

Get the data saved on our local disk

To mount and enable access to our Google Drive, we will execute the following code:

from google.colab import drive
drive.mount ('/ content / gdrive')

Google Colaboratory is a virtual machine environment based on Jupyter Notebooks.

They can be run in the cloud, it is possible to choose to run our notebook on a CPU, GPU, or a TPU for free. It has some restrictions, such as that a session lasts 12 hours, after that time our environment is cleaned and we lose the variables and files that we have stored there.

They are very convenient for beginners who want to experiment with machine learning and deep learning but without incurring cloud processing costs. In addition, the work environment already comes with many libraries installed ready to use, such as Tensorflow, saving us the work of setting up our development environment.

Cell types

As in Jupyter Notebook, we have code cells and markdown or text cells.

To insert cells, if we place the cursor between two cells, we get the option to add text or code cells as seen in the following image. We can also do it from the Insert menu, or from the more code and text buttons on the top bar.

An interesting feature of Colabs is the possibility of adding a temporary code cell. We insert it from the Insert menu, they are displayed as an auxiliary window in the browser as seen in the following image. It is very useful for quick tests without adding extra cells in our notebooks.

How to use datasets in Colab?

One of the first things we're going to want to do is use our datasets on the notebook. There are several ways to do it.

Import from Google Drive

A simple way to do this is by importing the files from Google Drive. In Colab we have code fragments available that solve this integration with Drive.

It is advisable to investigate the rest of the code snippets provided, several common file handling tasks and even visualizations are there solved.

Import from Github

A very convenient alternative is to import our datasets to Colab from a git repository hosted for example on Github (or Bitbucket). We can clone the repository to our workspace simply with the following line:

How to import your own functions from .py files

In order not to overload our notebook and reuse functions from one notebook to another, it is desirable to keep code with certain functions in separate Python files. In Jupyter Notebooks we usually import .py files with different helper functions, this is also possible in Colab.

I can perform it as shown in the following code snippet. A git repository must be cloned containing a file (filename.py in the example) with the functions I want to use.