Item — HN Reader

I now use chatGPT for translating english to traditional Chinese and give me some variations. Definitely works better than Google translate.

For example for "I want to build a good gym routine":

我想建立一個良好的健身計畫。 Wǒ xiǎng jiànlì yīgè liánghǎo de jiànshēn jìhuà.

Here are some variations of the sentence along with their English translations: 我想制定一套有效的健身計劃。 Wǒ xiǎng zhìdìng yī tào yǒuxiào de jiànshēn jìhuà. I want to create an effective workout plan.

我希望安排一個健康的健身日程。 Wǒ xīwàng ānpái yīgè jiànkāng de jiànshēn rìchéng. I hope to set up a healthy fitness schedule.

ps: I am building an app to make learning chinese easier. Feel free to ping me privately for testing :)

1 comments

I also do this with translations between Swedish, Danish and English.

I find it fascinating that chatGPT is better at translating (a lot better) than a tool specifically built for it, when (if I understand correctly) chatGPT was in no way designed to translate, and is only in some way predicting text, one word at a time.

How come google hasn't leveraged better existing tech to make the translations better? Is it too computationally expensive?

The Google translate offline translation datasets are absolutely tiny - like, 20mb in size for French. Obviously this is heavily quantised and so on, but it’s not that surprising that a 30-60+ billion parameter language model outcompetes Google translate handily.

I assume you’re right - that Google translate hasn’t been updated to take advantage of much bigger, more computationally complex models. I suppose for direct translation it’s probably not needed, but being able to ask chatgpt to explain the translation (and any cultural nuances involved) is a game changer when you’re trying to learn a language.

But they could easily be using a larger dataset online?

Which goes two ways: maybe this line of reasoning doesn't mean anything; or well yes exactly, but why so small online when they have all this space and also offer Bard.

They don't charge anywhere near enough to do that, I'd imagine; and likely couldn't at the scale they operate at (I mean, they are even embedded into many apps to help instantly translate banal things like comments). Imagine trying to translate a long news article with a sequence of max-length LLM inferences.

Yeah I'm not actually suggesting it run through Bard/a LLM, I just mean surely small dataset size is a design requirement for space constrained devices' offline translation, it doesn't necessarily mean they use the same datasets online, and if they do.. why, because it seems to be enough?

(It's a bit confusing to talk about because surely it is just an older version of the same sort of thing, it's a less large language model right? I just think it could/should/would be a bit larger in the online hosted version.)

You can check if translations are also better on Bard, that would partly answer that question

I assume the low latency you get from Google Translate is not feasible with current LLMs like ChatGPT. Translate is used to translate sentences on the go, live videos, translate entire web pages ; all of these would be too expensive (and slow) with an LLM... but things might change in the coming months/years as the tech improves.