Trying to bring my understanding of a few concepts and optimizations that have happened over the past few years in NLP to end boss level.

Disclaimer: No Vibe Coding - I read docs & papers and type every line of code myself (you can blame me on everything lol O_O)

• derived self-attention from scratch
• implemented my own version of the original transformers encoder-decoder architecture from "Attention is All You Need"
• built a proof of concept implementation of a byte pair encoder