As machine translation becomes available in more languages with MinT and NLLB-200 models, we want to expand support to a new set of languages. These languages have already machine translation support, but the feedback and/or data suggest that they could benefit from an additional option.
This ticket proposes to enable MinT using NLLB-200 as an option for the following languages:
- Afrikaans (af/afr_Latn) ✅
- Armenian (hy/hye_Armn) ✅
- Bengali (bn/ben_Beng) ✅
- Estonian (et/est_Latn) ✅
- Haitian Creole (ht/hat_Latn) ✅
- Hindi (hi/hin_Deva) ✅
- Irish (ga/gle_Latn) ✅
- Javanese (jv/jav_Latn) ✅
- Lithuanian (lt/lit_Latn) ✅
- Norwegian (no/nob_Latn) We want to pay special attention to verify there is no issue with alternative codes for Norwegian such as "nb", or with different variants like Norwegian Nynorsk (nn). Issue with codes captured in T339091 ✅
- Swahili (sw/swh_Latn) ✅
- Thai (th/tha_Thai) ✅
- Uyghur (ug/uig_Arab) ✅
Steps: