加载中…
加载中…
source language: pcm (Nigerian Pidgin)
target language: en (English)
dataset: Parallel Sentences from the Pidgin and message translations (English) of the Bible.
model: transformer-align
pre-processing: normalization + SentencePiece
| test set | BLEU |
|---|---|
| 20% of the bible data | 21.4 |