This folder contains the MorisienMT dataset.

Files:
1. train.$lang1-$lang2.$lang1 train.$lang1-$lang2.$lang2 are parallel corpora for $lang1 and $lang2. $lang1-$lang2 can be cr-en, en-cr, cr-fr and fr-cr where cr is Morisien, fr is French and en is English.
2. dev.$lang and test.$lang are 3 way development and test sets where $lang is en, cr and fr.
3. Creole MT evaluation Annotations.xlsx contain the human annotations of the 50 examples for all 4 translation directions for the NMT system we evaluated.