After several rounds of enablements of MinT for languages supported by NLLB-200 including those not supported by other services (T326578) and those with the mobile translation experience available (T339105), NLLB-200 is used to support 152 languages in Content and Section translation. However, NLLB-200 could support an additional set of Wikipedias.
This ticket proposes to enable MinT using NLLB-200 as an option for the following languages:
- Scottish Gaelic (gd/gla_Latn) — ✅
- Hungarian (hu/hun_Latn) — ✅
- Italian (it/ita_Latn) — ✅
- Kazakh (kk/kaz_Cyrl) — ✅
- Kyrgyz (ky/kir_Cyrl) — ✅
- Minangkabau (min/min_Latn) — ✅
- Dutch (nl/nld_Latn) — ✅
- Polish (pl/pol_Latn) — ✅
- Sardinian (sc/srd_Latn) — ✅
- Swedish (sv/swe_Latn) — ✅
- Ukrainian (uk/ukr_Cyrl) — ✅
- Vietnamese (vi/vie_Latn) — ✅
- Finnish (fi/fin_Latn) — ✅ Another ticket (T333969) will add support using OPusMT for specific pairs, but enabling NLLB-200 will be useful as a fallback model for all language pair combinations.
- Simple English (simple/eng_Latn) — ✅ Already enabled
Japanese (ja/jpn_Jpan)German (de/deu_Latn)English (en/eng_Latn)
Ideally, all languages listed in NLLB-200 documentation should be available for Content Translation with MinT, with the exceptions of languages where MT has not been enabled yet (English and German), requested to be disabled (Japanese in T323973), and those languages with no Wikipedias yet (T336683).
Steps:
- Enable MinT for the selected languages.
- Communicate with the communities.
- Scottish Gaelic
- Hungarian
- Italian
- Kazakh
- Kyrgyz
- Minangkabau
- Dutch
- Polish
- Sardinian
- Swedish
- Ukrainian
- Vietnamese
- Simple English