If you want to jump right into training complex models and become a serious deep learning practitioner, there will be one thing that you’ll need: Hardware. Specifically, a powerful GPU (Graphics Processing Unit) is what you would want_. _Trust me, I’ve never seen someone with a straight mind train a decently-complex CNN only using a CPU. However, unless you have a lot of cash in your pocket, having access to even one GPU is not going to be easy nor inexpensive, let alone a GPU that has enough computing power or memory for your need. And let’s be serious, we ALL wish to have all the money in the world to dish out multiple RTX 2080 Ti’s just for our own needs.
What if you are in a tight budget but want to pursue deep learning? Heck, what will you do if you have no cash at the moment? Don’t worry: there are places to get started for free, where you can begin model experimentations and have access (albeit limited) to a fully-fledged GPU. If you’re dead stuck on where to begin with your pockets empty, here are some options you can choose, all of which will provide you a Jupyter (notebook or lab) environment to start creating your projects.
How will it work?_ Some providers will store a copy of your custom environment into their servers (a.k.a. the cloud). Others always load a “one-fits-all” environment during every launch while making it very convenient to install additional packages._
Before we begin, here are some notable trends among these free providers that you should definitely keep in mind:
Without further ado, here are some of the best free options for a GPU instance to kickstart your deep learning experiments!
NOTE: Since all of the choices below offer a pretty decent GPU to start with, I’ll be focusing more on the features offered by each, than the GPU performance and stats.
#programming #data-science #data analysis
Nor do you need any hardware, and it’s all thanks to the “Cloud”.If you want to jump right into training complex models and become a serious deep learning practitioner.