Large language model: Can you say less? What on earth do you want to do?

Large Language Model: Can you talk less? What on earth are you trying to do?

via cnBeta.COM Chinese Industry Information Website - Telegram Channel

Telegraph
Large Language Model: Can you talk less? What on earth are you trying to do?

Large Language Model: Can you talk less? What on earth are you trying to do? Yes, it's about hot pot. However, these once magical prompt words that made you more efficient may have become obsolete. Let's put it this way, without realizing it, large language models have actually been divided into two camps: traditional general large language models and reasoning large language models. For example, GPT-o1 is not a direct version upgrade of GPT-4o. 4o belongs to the general large language model, while o1 is already a reasoning model. Similarly, the V3 version that DeepSeek uses by default is a general large language model, and only when you click the "Deep Thinking" button in the lower left corner will you use the R1 reasoning model. In the era of reasoning models, the more detailed the prompt words are, the stupider the AI may become. For example, in OpenAI's official documentation, under the column of reasoning models, it clearly states that used...