Collocates multiple machine translation outputs for qualitative analysis. There's a lot of additional flags to make life easier.
python3 mt_meld.py --src source.txt --ref reference.txt --hyps opennmt.out nematus.out fairseq.out
Src: Esto es una prueba
Ref: This is a test
MT1: :-) This is a test
MT2: That was a dog
MT3: :-) This is a test
...
--google en
Argument is the target language code
--del_bpe
--truecase target_truecase_model.txt
--detok en
Argument is the target language code
--lc
--head 5