translations

NMT is a category containing both transformers and deep RNN. The Mozilla translation models are transformer LLM NMTs trained via Marian https://marian-nmt.github.io/ (ref: https://github.com/mozilla/translations/blob/main/docs/READM...)

NMT is a category containing both transformers and deep RNN. The Mozilla translation models are transformer LLM NMTs trained via Marian https://marian-nmt.github.io/ (ref: https://github.com/mozilla/translations/blob/main/docs/READM...)

No, EU funded Bergamot and HPLT projects. Integration came much later, but again as a part of Bergamot project. [0]: https://browser.mt/ [1]: https://hplt-project.org/ [2]: https://github.com/mozilla/translations?tab=readme-ov-file#a...

I'm encouraged that they're actively exploring this and not shying away from experiments. It seems clear to me that there are areas of Mozilla[0],[1] pushing closer and closer to great local AI integrations doing the kinds of things that I, a browser user, find useful. I went to some of the articles on the hn frontpage and had questions (and followups!) that begat reasonable starting points for further learning. Hopefully they continue to iterate on this with better integration (for instance, moving to a toolbar icon instead of persistent badge on every page) and then make it ~truly privacy respecting by moving locally. [0] https://github.com/Mozilla-Ocho/llamafile [1] https://github.com/mozilla/translations

Thanks for using it! Best way currently is to keep using and reporting issues on [1]. You can see how the models are trained on [2] and file issues there too. [1] https://github.com/mozilla/firefox-translations/issues [2] https://github.com/mozilla/firefox-translations-training

All of them are freely available. Most of them through mtdata [1]. The exact list of the datasets is in the firefox-translations-training pipeline configuration file [2]. [1] https://pypi.org/project/mtdata/ [2] https://github.com/mozilla/firefox-translations-training/blo...

The training pipeline is also on Github! [1] I was experimenting with running the wasm version of bergamot-translator (the translation engine used by the addon) in node [2]. However, if you want more performance, using the Python library [3] or the native C++ interface [4] gets you further because the wasm build is limited to a single thread and thus a blocking interface, and can't use all the processor specific optimisations that are in the native builds. EDIT: Another option is using translateLocally [5], which is a Qt desktop app based on bergamot-translator. It has a native messaging API that is designed as a much faster alternative to the wasm build for browser extensions, but it can also be used from Python [6]. [1] https://github.com/mozilla/firefox-translations-training [2] https://gist.github.com/jelmervdl/a4c8b6b92ad88a885e1cbd5

Is this open source? I don't see a github link anywhere, and I'm not sure if the models are freely usable. EDIT: maybe this is it: https://github.com/mozilla/firefox-translations-models also some info here: https://github.com/mozilla/firefox-translations-training