3 ms·
Another plausible alternative is that Telegram embeds an open source NLP model in the app so translation can be performed locally (offline). To be fair we know
by criticaltinker 5y ago
Another plausible alternative is that Telegram embeds an open source NLP model in the app so translation can be performed locally (offline).
To be fair we know that mobile hardware resources can be extremely limited, and I’d wager a server side model will always be bigger and therefore better.
But recently there’s been a lot of amazing progress in techniques to shrink large models down while preserving most of the accuracy (eg quantization aware training, etc).
With some additional constraints on scope (maybe only supporting the handful of languages the user needs?), I believe a sharp team of a few experts could deliver this fairly quickly, with reasonable results that would of course improve over time.
- mikaraento 5y agoGoogle also provides an official on-device translation SDK that is free-as-in-beer (https://developers.google.com/ml-kit/language/translation https://developers.google.com/ml-kit/language/translation). The models are around 20M for any single language so there is a not insignificant cost on the user in terms of delay for downloading, data usage and disk space. Disclaimer: I work for Google and worked on the ML Kit Translate SDK, but I don't speak for Google.
- freemint 5y agoIs integration with some Google Services App required or do those models also run on de-googled Android distributions?
- mikaraento 5y agoML Kit only runs on devices with Play Services, sorry.