firefox-translations-models

The local translation in Firefox is an NMT model, not an LLM: https://github.com/mozilla/firefox-translations-models?tab=r... It's much more efficient on system resources than the larger LLMs downloaded by browsers for other tasks.

It is AI-based, you can check the source code: https://github.com/mozilla/firefox-translations-models

The models[0] are open and great, I find them so useful that I ended up making an Android app[1] just to have them available also outside of Firefox. [0]: https://github.com/mozilla/firefox-translations-models/ [1]: https://github.com/davidventura/firefox-translator

You can use bergamot ( https://github.com/browsermt/bergamot-translator ) with Mozilla's models ( https://github.com/mozilla/firefox-translations-models ). Not the easiest, but easy enough (requires building). I used these two projects to build an on-device translator for Android.

Well there's this[0] but unfortunately I don't see any results and I'm not about to put in the time to run it myself. Responding to GP, I won't object that LLMs aren't optimized for translation but I would generally expect them to perform quite well at it given that it seems to come with the territory of being able to respond to natural language prompts in multiple languages. > massive, overfit generative models are awful at translation > part-of-speech classification + dictionary lookup + grammar mapping – an incredibly simplistic system with performance measurable in chapters per second – does a better job Those are two distinct claims you made and I'm not inclined to accept either of them without evidence given how unexpected both of them would be from my perspective. [0] https://github.com/mozilla/firefox-translations-models/tree/...

Maybe you can develop it, if you want. The models and the source code for running it are all there, someone just have to wrap it in a standalone binary instead of a WebExtension. - https://github.com/mozilla/firefox-translations - https://github.com/mozilla/firefox-translations-models - https://github.com/mozilla/bergamot-translator Note that most of the heavy NLP code used (MarianMMT, fasttext) were originally written as C++ libraries but were compiled to WebAssembly to make a WebExtension out of it. So if you want to make a command line / GUI desktop app there are two choices: use C++ directly if you want to make an app with minimal dependencies, or use JS/WebAssembly to make a NodeJS / Electron based app (which might be heavier but more accessible for JS devleopers.)

Looks like a french model was added 17 days ago https://github.com/mozilla/firefox-translations-models/tree/...

Yes, the extension is being developed here [1] and the engine[2]/wasm[3] wrapper here. The models and training pipeline are in the urls your posted and the evaluation of those models are hosted here [4] [1] https://github.com/mozilla/firefox-translations [2] https://github.com/browsermt/bergamot-translator [3] https://github.com/browsermt/bergamot-translator/tree/main/w... [4] https://github.com/mozilla/firefox-translations-models/tree/...

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