This! The best resource I've found to explain transformers, that made them clear to me. I wish all deep learning papers were written like this, using pseudocode.