Token Level Quality Estimation on FCE (test)
81.85PrecisionBERT-VERNet
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| BERT-VERNetScenario=Hypothesis, Input Encoding=Source, hypothesis pairs2021.05 | 81.85 | 44.27 | 69.97 | |
| BERT-VERNetScenario=Source, Input Encoding=Source, hypothesis pairs2021.05 | 81.53 | 45.71 | 70.48 | |
| ELECTRA-VERNetScenario=Source, Input Encoding=Source, hypothesis pairs2021.05 | 80.94 | 50.51 | 72.24 | |
| ELECTRA-VERNetScenario=Hypothesis, Input Encoding=Source, hypothesis pairs2021.05 | 80.62 | 49.16 | 71.48 | |
| BERT-GED (HYP)Scenario=Hypothesis, Input Encoding=Source, hypothesis pairs2021.05 | 80.27 | 40.58 | 67.14 | |
| BERT-GED (JOINT)Scenario=Hypothesis, Input Encoding=Source, hypothesis pairs2021.05 | 76.71 | 46.94 | 68.07 | |
| BERT-GED (JOINT)Scenario=Source, Input Encoding=Source, hypothesis pairs2021.05 | 75.62 | 44.44 | 66.32 | |
| BERT-GED (SRC)Scenario=Source, Input Encoding=Source sentences only2021.05 | 74.22 | 43.34 | 64.97 |