Sentence Level Quality Estimation on CoNLL 2014 M2 (test)
69.97PrecisionELECTRA-VERNet
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ELECTRA-VERNetbackbone=ELECTRA2021.05 | 69.97 | 42.12 | 61.8 | 37.18 | 28.77 | |
| BERT-GED (JOINT)backbone=BERT, input=Joint Source + Hypothesis2021.05 | 69.33 | 36.02 | 58.51 | 28.62 | 16.28 | |
| BERT-GED (HYP)backbone=BERT, input=Hypothesis2021.05 | 68.9 | 34.35 | 57.36 | 30.06 | 16.79 | |
| BERT-VERNetbackbone=BERT2021.05 | 68.75 | 40.26 | 60.22 | 31.02 | 22.75 | |
| NQE (CR)variant=CR2021.05 | 62.52 | 35.24 | 54.14 | 24.8 | 9.12 | |
| BERT-QEbackbone=BERT2021.05 | 62.24 | 38.27 | 55.31 | 22.85 | 12.17 | |
| NQE (RR)variant=RR2021.05 | 61.38 | 33.03 | 52.39 | 23.43 | 6.62 | |
| NQE (CC)variant=CC2021.05 | 60.62 | 35.77 | 53.23 | 22.94 | 8.39 | |
| NQE (RC)variant=RC2021.05 | 60.09 | 33.11 | 51.67 | 24.12 | 5.52 | |
| BERT-GED (SRC)backbone=BERT, input=Source only2021.05 | 52.98 | 52.07 | 52.79 | 3.78 | 20.56 | |
| BERT-LMbackbone=BERT2021.05 | 52.82 | 49.59 | 52.14 | 3.47 | 17.62 | |
| BERT-GQEbackbone=BERT2021.05 | 52.67 | 50.39 | 52.19 | 2.56 | 14.54 |