Hacker News Reader

Eh. I enjoy the variety of languages of GT, but the quality is (imo) better with DeepL[1], and Yandex Translate[2] has some minor languages of Russia and ex-USSR which either aren't supported by Google or aren't supported on the same level. E.g. Bashkir and Mari. GT is still very impressive, don't get me wrong. But it has some strong competitors.

[1] https://www.deepl.com/translator

[2] https://translate.yandex.com/

Thanks, didn't know good alternatives existed.

Google Translate produces mediocre results with European languages and absolute gibberish with Chinese or Japanese.

Personally, Google Translate is my litmus test for "is ML actually any good". If we ever reach the point of fluent translation of German or coherent translation of Japanese, I will be extremely impressed.

Automatically producing mediocre results is amazing.

So: Google Translate IS amazing. The result of Google Translate is definitely not amazing, but that wasn't the claim.

You mean a neural network I could pull off github, and if I had enough money to train it, end up with the same results?

You do know that they designed the NN they trained as well as collected the data set, right?

We aren't to the place where a single individual with enough compute resources could feasibly replicate google translate.