Running 134 Natural Language Processing before LLMs ๐ 134 From n-gram counts to attention, learning by doing