Neural Machine Translation versus Large Language Models / hacker news