According to this tutorial [1] by Google, part of why LLMs are so verbose is a phenomenon called 'chain of thought reasoning'. Basically, the LLM will formulat

According to this tutorial [1] by Google, part of why LLMs are so verbose is a phenomenon called 'chain of thought reasoning'. Basically, the LLM will formulate a better answer to the question if it talks itself through its reasoning process. [1] https://youtu.be/zizonToFXDs?si=5f_IxvR7h0iJy2Db&t=678 https://youtu.be/zizonToFXDs?si=5f_IxvR7h0iJy2Db&t=678