List your company
ResearchFormerly WizWand

Machine Translation Evaluation on WMT17 (test)

0.653Kendall TauParaBLEU
0.277560.375030.47250.56997Jul 17, 2021
Updated 8mo ago

Evaluation results

MethodLinks
ParaBLEUbackbone=RoBERTa-large, size=large, layers=24, hidden_units=1024, heads=16, pretraining_epochs=42021.070.6530.843
BLEURT-largetype=learned neural metric2021.070.6250.818
ParaBLEUbackbone=RoBERTa-base, size=base, layers=12, hidden_units=768, heads=12, pretraining_epochs=42021.070.5890.785
BERTScorebackbone=DeBERTa-large, type=non-learned neural metric2021.070.580.773
BERTScorebackbone=RoBERTa-large, type=non-learned neural metric2021.070.5670.759
BERTScorebackbone=T5-large, type=non-learned neural metric2021.070.5360.738
chrF++type=character-level n-gram-matching2021.070.3960.578
ROUGEtype=n-gram-based2021.070.3540.518
TERtype=edit-distance-based2021.070.3520.475
MoverScoretype=non-learned neural metric, algorithm=optimal transport2021.070.3220.454
METEORtype=n-gram-based2021.070.3010.443
BLEUtype=n-gram-based2021.070.2920.423