Comments by "T J" (@TJ-hs1qm) on "[1hr Talk] Intro to Large Language Models" video.

  1. There's no sematic tree . Only linear algebra and reinforced learning. Meaning cheap workers slaveing off in Kenia for 1 $ / h making a bunch of billionaires richer
    1
  2. ​ @smkh2890 he has more in depth tutorials on his channel you can look up word embeddings, transformer architecture, back propagation and gradient descent there's plenty everywhere. At the end it all comes down to linear algebra and some calculus. The magic happens due to the huge amount of training data.
    1