Page MenuHomePhabricator

Check and document translation samples for language pairs supported by Softcatalà and NLLB-200
Open, MediumPublic

Description

Softcatalà translation models were requested to be integrated into Content Translation (T284905). An initial language pair (English to Catalan) was integrated as part of MinT, but Softcatalà supports the following language pairs:

  1. German (de) ↔ Catalan (ca)
  2. English (en) ↔ Catalan (ca)
  3. French (fr) ↔ Catalan (ca)
  4. Galician (gl) ↔ Catalan (ca)
  5. Italian (it) ↔ Catalan (ca)
  6. Japanese (ja) ↔ Catalan (ca)
  7. Dutch (nl) ↔ Catalan (ca)
  8. Occitan (oc) ↔ Catalan (ca)
  9. Portuguese (pt) ↔ Catalan (ca)
  10. Spanish (es) ↔ Catalan (ca)

Since these language pairs are also supported by NLLB-200, we want to compile some samples and check which language pairs seem to be working better with each model.
Changes can be proposed as part of the updates to be done after re-running the report on machine translation service usage (T338606). In this way, the next run of the report will allow to compare data about translation modifications and deletions for those pairs.

Event Timeline