All in all, this makes your work easier to understand at a glance—whether it's purely text or something more visual.

总而言之，这会让你的作品更能够眼就清楚明了－－不管是或是更可视化的东西。

Traditionally, multimodal models are created by stitching together text-only, vision-only, and audio-only models in a suboptimal way at a secondary stage.

传统技术来说，多模态模型是通过第二阶段以次优方式将、视觉和音频模型拼接起来创建的。

Microsoft said anyone can describe a calculation they would like in plain text and its AI program would create a spreadsheet.

微软称，任何人都可以描述他们中想要的计算，其AI程序将创建出相应的电子表格。

Plain text spam emails have a carbon footprint of 0.3 grams of CO2 emissions.

垃圾邮件的碳足迹为 0.3 克二氧化碳排放量。

And this is just pure text, right?

这只是，对吧？

The input, we're going to start calling plaintext.

输入，我们将开始调用。

It can be in the form of text, photos or text with photos.

可以是、图片形式， 或是图结合。

Side note, when you make an ad on LinkedIn, can it be a text only ad?

顺便提，领英上做广告时，可以只用广告吗？

What happens, they ask, if the translated plaintext is very short, but says " We will make your sun go supernova tomorrow" ?

他们问，如果翻译出的很短，但内容是" 明天我们会让你们的太阳变成快爆炸的超新星" ，会发生什么？

And so what I didn't expect was just how good a model could get training purely on text.

而且我没想到的是，仅仅通过训练，个模型能够变得多么好。

When it emerged towards the end of the 1980s as a purely text-based medium, it was seen as a tool to pursue knowledge, not pleasure.

20 世纪 80 年代末当它作为种媒介出现时，它被视为种追求知识而非快乐的工具。

" Currently, most linguistic datasets remain limited to single-text formats and specific academic domains, lacking the scale and diversity required for AI applications, " Wang said.

王晖指出，当前语言数据资源仍主要集中于形态与特定学术研究领域，数据规模与类型多样性方面存明显不足，难以满足人工智能技术发展的多维需求。

Because plain text files don't have a way to specify what's a link and what's not, it was necessary to develop a way to " mark up" a text file with hypertext elements.

因为件没有指定什么是链接、什么不是链接的方法，所以需要开发种用超元素来“标记”件的方式。

To perform this, of course, we need a lot of training data, and here, the AI reads 40 gigabytes of internet text, which is 40 gigs of non-binary plaintext data, which is a stupendously large amount of text.

当然，为了达到目的，我们需要大量的训练数据，章里的人工智能已经读取了400亿字节的互联网，包括40GB的非二进制数据，这是极大的量。