Deep learning on Crestle

While a customized deep learning environment can be of use when you need greater control over the system—such as when you want to have third-party applications working along with your deep learning model—at other times, you may not have such needs, and you'll only be interested in performing deep learning on the cloud, quickly and in a collaborative manner. In such circumstances, paying the cost of an AWS g2.2xlarge instance would be much higher than that of paying only for computing time or GPU time used.

Crestle is a service that provides GPU-enabled Jupyter Notebooks online at very affordable pricing. To begin using Crestle, go through the following steps:

1. Log on to www.crestle.com.
2. Click on Sign Up and fill up the sign-up form that appears.
3. Check your email for an account confirmation link. Activate your account and sign in.
4. You'll be taken to the dashboard where you'll find a button reading Start Jupyter. You will have the option of using the GPU or keeping it disabled. Click on the Start Jupyter button with the GPU option enabled.

You will be presented with a Jupyter environment running on the cloud with GPU support. While the pricing is subject to change with the passage of time, it is one of the most affordable solutions available on the internet as of January 2020.