I’m giving a seminar series on how LLMs work and the math that powers them. This is a teaser for the first video. You can sign up here.

The agenda is as follows:

• Session 1: Neural Networks, their structure, matrix representation, feed forward processing and back propagation.
• Session 2: Embeddings, how they are computed and different ways you can use them.
• Session 3: Attention, how it works, the matrix math behind it and how it solves the context problem with embeddings.
• Session 4: An overview of the Transformer, how it works, how it’s trained and how it produces such uncanny results.