Code Scare

Documentation for cloud computing platforms such as Google Cloud, AWS or Azure can sometimes be hard to understand. I’ve tried to enlist a few features and explain them in short below. Compute Instances These are virtual machines running on physical servers in whichever region you choose running the OS of your choice (generally GNU/Linux). Since these are virtualized (or in some cases containerized), they run on vCPUs (virtual CPUs), which are slower than the hardware powering them, but fast enough nonetheless.

If you’re using Tensorflow to train a deep learning model that takes hours to run on good hardware, chances are that you’ve heard of Google’s latest cloud offering - the Tensor Processing Unit (TPU). According to Google, these ASICs can be 15x to 30x faster than the hardware you are currently using. However, if you glance at the documentation for using cloud TPUs with TensorFlow, you will find that Google always refers you to use the Estimator API, which is a high level API built on top of the lower level Graph and Session APIs.

UPDATE - Support is officially here!! Background GitHub Pages supports SSL for pages with username.github.io, but till now, there was no inherent support for SSL for custom domains. As soon as you added a custom domain, your site would be served over plain HTTP. Enter ‘Enforce HTTPS’ Recently, GitHub added another option to the settings of repositories with GitHub Pages enabled, named Enforce HTTPS. As the documentation states, this is applicable only for github.

I previously wrote on setting up Hugo on GitHub Pages, but if you want to get slightly more ambitious and automate deployment to GitHub Pages, read on! At the end of this post, you will never need to build before commiting again; just write in markdown and let Travis CI take care of the rest! A note: This blog is actually doing this, so if you want a live example of the result of this procedure, head over to this GitHub repository.

If you are working on a RESTful API, chances are that you have read or heard (I’ll explain soon how it is pronounced) the word HATEOAS sometime while browsing best practices, technologies or about APIs in general. Wikipedia defines it as Hypermedia As The Engine Of Application State (HATEOAS) is a constraint of the REST application architecture that distinguishes it from other network application architectures. but (in my opinion), it doesn’t really explain much about how you are supposed to use it, or what the advantages (and disadvantages, if any) of using it are.

At the time of writing this, this blog itself is hosted on GitHub Pages and is built with the static site generator Hugo, and since this is just the second post, I’ll go into some detail explaining how to setup Hugo properly (or at least the way I did it), which should be a matter of minutes even if you are not so experienced with these sort of things! If you follow the steps exactly, you’ll have a blog up and running in no time!

Yup! This is indeed my very first post! I think its really intresting that you stumbled upon this, because I doubt it is going to be of much use to you! Nevertheless, I’m currently setting up this blog in Hugo, so everything here is in markdown, which I’m starting to really love! I guess I’ll make a separate post about my experience of setting up Hugo; maybe that will be more useful.