Comment by abusaidm

Nice write up Sebastian, looking forward to the book. There are lots of details on the LLM and how it’s composed, would be great if you can expand on how Llama and OpenAI could be cleaning and structuring their training data given it seems this is where the battle is heading in the long run.