Machine Translation on newstest En-Cs 2013 (dev)
0.1983BLEUBi-scale Character-level NMT (Row j)
Evaluation results
| Method | Links | |
|---|---|---|
| Bi-scale Character-level NMT (Row j)Target=Char, Depth=2, Attention=Concatenated, Model=Bi-S, Ensemble=true2016.03 | 0.1983 | |
| Base Character-level NMT (Row i)Target=Char, Depth=2, Attention=Concatenated, Model=Base, Ensemble=true2016.03 | 0.1952 | |
| Base BPE-level NMT (Row h)Target=BPE, Depth=2, Attention=Concatenated, Model=Base, Ensemble=true2016.03 | 0.1921 | |
| Base Character-level NMT (Row i)Target=Char, Depth=2, Attention=Concatenated, Model=Base, Ensemble=false2016.03 | 0.1768 | |
| Bi-scale Character-level NMT (Row j)Target=Char, Depth=2, Attention=Concatenated, Model=Bi-S, Ensemble=false2016.03 | 0.1762 | |
| Base BPE-level NMT (Row h)Target=BPE, Depth=2, Attention=Concatenated, Model=Base, Ensemble=false2016.03 | 0.1612 |