Google colab should be a very good choice. It will be free for a long time. We will never be soft on the wool of capitalism. The following is a very simple three-step tutorial to get started.
1. Configuration
Its configuration is about 15g hard disk (the cloud space of Google drive). Of course, various library files are not included. You can understand it as a data storage space. The memory is about 12gb. There are currently 4 types of gpu, p100 , T4, P4, K80, built-in complete library files of tensorflow and pytorch, if you want to use other deep learning libraries, you can do it with one pip, which can basically meet the experience of most projects or competitions.
For the use of colab, it only takes three steps to get started easily:
Prerequisites before use:
A tool that can be used on Google. The tools here generally include billing by time and billing by traffic (of course there are also various free ones). It is more cost-effective to recommend billing by traffic.
2. Steps
- Create a colab file
- Website https://drive.google.com/drive/my-drive
In terms of use, click New in the previous picture, then click More, and then click colab to create a new file. Of course, if it is the first time to use, the colab option may not be displayed, just click to associate more applications.
It is consistent with the usage of jupyter
2. Configure the GPU environment
Click the code execution tool, choose to change the runtime type, choose GPU, and save it!
You can also connect to the goole cloud disk.
Click the red dot on the picture,
run this code,
from google.colab import drive
drive.mount('/content/drive')
Now the configuration is all done
3. Skillful operation
