4 ms·
Protip: these magic Rosetta stones you hold in such high regard are trained on corpora of texts translated by professional translators with degrees in translati
by otabdeveloper4 3y ago
Protip: these magic Rosetta stones you hold in such high regard are trained on corpora of texts translated by professional translators with degrees in translation.
If you think these corpora we have today are enough to power the ML forever and ever until the end of time, then you should have another and better thought instead.
- mufti_menk 3y agoAs the ML improves the algorithms themselves can be used to generate more training data.
- otabdeveloper4 3y agoNo, they cannot. (At least not in any meaningful sense.) This will not create new information you can draw upon to improve translations models and keep them up to date.
- mufti_menk 3y agoThe model can generate translations and humans can verify that they're accurate enough to be used for training, making corrections if needed. Verifying work is much faster than creating it.
- otabdeveloper4 3y agoa) You will definitely need trained and credentialled translators to do the verifying. b) The claim that verifying is faster than creating is unsupported. (E.g. reviewing programmer code is tougher than writing from scratch.)
- mufti_menk 3y agoTry reading a book vs writing one
- trey-jones 3y agoBelieve it or not, the First Law of Thermodynamics seems to apply here!
- spywaregorilla 3y agoThat's not correct. Consider a human who can speak both spanish and english. They can translate intrinsically. Why? Because language is an interface to their knowledge. You do not require a special intermediary language that only trained professionals can teach. Language -> knowledge (embeddings) -> Language is fine
- otabdeveloper4 3y ago> They can translate intrinsically. Unfortunately no. If you speak two languages natively you know how hard translating can be unless you have training.