Large Language Models Explained

This post was published in the Common Craft Newsletter. Subscribe here.

Using a tool like ChatGPT can feel like chatting with an omniscient human. It appears to know virtually everything and is able to communicate clearly. It can understand what we want with minimal input.

You may wonder: Does ChatGPT learn and understand all the subjects? Does it truly know anything?

The answer to these questions lies in something called a Large Language Model or LLM. The model is created by computers that attempt to capture every word on the internet; around 1 trillion words. Then, massive computers analyze the context and connections between words and use that information to respond to chat requests.

A big part of understanding ChatGPT means understanding LLMs. And that was my challenge. I set out to explain LLMs in a two-minute video with a couple of constraints:

  1. It needed to be a Readable Video
  2. The video needed to work as an animated GIF

Large Language Models Explained

The video below is available on ​YouTube​.

Video Workaround for Email

One of the problems with online videos is that they don't play in most email programs. We're forced to link to a website in order to see the video play. And of course, most people don't click the links.

But all is not lost. Animated GIFs DO play in email. It's possible to include an animated gif that explains an idea right in the body of the message. For this to work, the GIF needs to be small in file size and consumable without sound.

This is a perfect situation for ​Readable Videos. I converted the video above to an animated gif that's two minutes long and under 2mbs in size. I converted the video above into a GIF using the free tool ​gifski​.

Animated GIFs aren't perfect for every situation, but they offer a way to share an idea in the form of an animated video. This is one of the many skills we teach in the ​Readable Videos course​.