Fine-tuning NLLB-200 for a New Low-Resource Language in 2026
• 2
FormosanBank
non-profit
AI & ML interests
None defined yet.
Recent Activity
View all activity
Articles
Current NLLB-200 SPM8k directional models for 15 Formosan languages with English and Traditional Chinese.
-
FormosanBank/nllb200-formosan-en-spm8k
Translation • 0.6B • Updated • 28 -
FormosanBank/nllb200-en-formosan-spm8k
Translation • 0.6B • Updated • 31 -
FormosanBank/nllb200-zh-formosan-spm8k
Translation • 0.6B • Updated • 30 -
FormosanBank/nllb200-formosan-zh-spm8k
Translation • 0.6B • Updated • 37
Best ASR checkpoint per Indigenous Formosan language, selected on TACL dataset_v1/split_v2 all-corpora evaluation.
-
FormosanBank/formosan-asr-amis
Automatic Speech Recognition • 0.3B • Updated • 29 -
FormosanBank/formosan-asr-atayal
Automatic Speech Recognition • 0.2B • Updated • 22 -
FormosanBank/formosan-asr-bunun
Automatic Speech Recognition • 0.3B • Updated • 24 -
FormosanBank/formosan-asr-kanakanavu
Automatic Speech Recognition • 0.3B • Updated • 21
Current NLLB-200 SPM8k directional models for 15 Formosan languages with English and Traditional Chinese.
-
FormosanBank/nllb200-formosan-en-spm8k
Translation • 0.6B • Updated • 28 -
FormosanBank/nllb200-en-formosan-spm8k
Translation • 0.6B • Updated • 31 -
FormosanBank/nllb200-zh-formosan-spm8k
Translation • 0.6B • Updated • 30 -
FormosanBank/nllb200-formosan-zh-spm8k
Translation • 0.6B • Updated • 37
Parallel Corpora for FormosanBank
Best ASR checkpoint per Indigenous Formosan language, selected on TACL dataset_v1/split_v2 all-corpora evaluation.
-
FormosanBank/formosan-asr-amis
Automatic Speech Recognition • 0.3B • Updated • 29 -
FormosanBank/formosan-asr-atayal
Automatic Speech Recognition • 0.2B • Updated • 22 -
FormosanBank/formosan-asr-bunun
Automatic Speech Recognition • 0.3B • Updated • 24 -
FormosanBank/formosan-asr-kanakanavu
Automatic Speech Recognition • 0.3B • Updated • 21
Canonical public audio for XML corpora published in FormosanBank. Each audio dataset uses the same license as its published XML.