Learn directly from spreadsheets-are-all-you-need-ai

This was a good introduction on how transformer models work. As someone who flunked calculus three times, it was good to have someone walk me through the scary math to get an intuitive understanding of LLMs. I may not be able to build an LLM from scratch after this course, but I have a much better understanding of how they work.

I've taken a few courses on ML, Computer Vision and NLP via Coursera and DeepLearning.ai. This class is very high. Ishan does a great job of taking a tough concept, understanding GPT2, and decomposing the model into components. Very clever how he used excel to simplify the understanding of the math of the model as well as additional spreadsheets to help understand the choice of functions (e.g. Softmax) used in the model. Cannot believe we covered so much in 5 classes. Also loved the context with AI research, e.g. Word2Vec for embeddings, "All you need is Attention!" paper to understand attention and the many other research paper references. Well done.

Helped me really get a concrete understanding of what is going on under the hood in a GPT model. You can stare at structure diagrams all day, or read a lot of hand-wavy explanations online, etc or you can sign up to this course and actually understand the matrix multiplies that are going on when you type in a query and get a response from ChatGPT. Awesome course.

Ishan is a master educator, deeply interested in this topic, and constantly hunting to better ways to communicate his understanding of this topic in easily understandable peices. A delight to experience that lightbulb moment of understanding just how the transformer architecture works, end to end.

A great course for anyone interested in LLMs. Ishan breaks things down in such a digestible and accessible way. I love being able to scroll through the web app and understand what each block is doing architecturally. This class pulled back the curtain on the “magic” of LLMs for me.