Beyond the hype, LLMs are useful for many things, particularly for knowledge workers. I use it daily for information searching, learning new topics, writing reports, and grammar correction.
It’s essential to understand the technology under the hood so we can know what’s possible and what’s limited.
Here are the materials that I found super useful to learn the fundamentals.
Deep Dive into LLMs like ChatGPT & How I use LLMs by Andrej Karpathy
In the AI circle, Karpathy is a very respected teacher and engineer. These two videos are useful to understand LLMs, no math or coding required. It’s understandable for both non-tech people and tech people.
Three episodes to explain LLMs by 3Blue1Brown
3blue1brown is a great teacher in math. His videos explained the complex math behind the LLMs with great visual effects, which make it much easier to understand. It requires a bit of linear algebra background. But even without it, the first episode Transformers, the tech behind LLMs, and the second episode Attention in transformers, step-by-step are useful to understand how LLMs “understand”.
One additional video from 3Blue1brown on neural networks is a must-watch to understand the very basics of AI.