Running 137 Natural Language Processing before LLMs ๐ 137 From n-gram counts to attention, learning by doing