Karpathy：和 LLM 语音聊十分钟，把混乱想法说清楚

中文译文

我发现，与 LLM 协作时，一个很有用的做法是来一场长时间的碎碎念。

有时，LLM 需要更多信息才能理解你想做什么，但你又懒得把这些信息打出来。这种时候，我喜欢往后一靠，切到 /voice，随口絮叨大约 10 分钟。内容可以完全不成章法，想到什么说什么，纯粹的意识流。

有时我会先声明一句，比如「我现在切到语音识别，出现拼写错误的话先说声抱歉……」。有时我会把它变成一场来回几轮的小型访谈。

但我发现，LLM 不知怎么的很擅长重建这种漫长、缺乏条理的絮叨。它常常会把你那团缠在一起的思绪整理得比你一开始说出来的清楚得多。

这样一来，你和 LLM 的配合会更顺，后面需要纠正它的地方也会变少。

英文原文

One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and often their echo of your own tangle of thoughts comes out quite a bit cleaner than what you started with. The result is that you improve the mind meld and have to correct things less from that point on.