| relation-classification-on-tacred | BERT | #4 | F1: 66.0 |
| relation-classification-on-tacred | ERNIE | #6 | F1: 68.0 |
| paraphrase-identification-on-quora-question-pairs | ERNIE | #15 | F1: 71.2 |
| semantic-textual-similarity-on-mrpc | ERNIE | #16 | Accuracy: 88.2% |
| semantic-textual-similarity-on-sts-benchmark | ERNIE | #20 | Pearson Correlation: 0.832 |
| relation-extraction-on-tacred | ERNIE | #23 | F1: 67.97 |
| natural-language-inference-on-multinli | ERNIE | #29 | Matched: 84.0Mismatched: 83.2 |
| natural-language-inference-on-qnli | ERNIE | #29 | Accuracy: 91.3% |
| linguistic-acceptability-on-cola | ERNIE | #30 | Accuracy: 52.3% |
| sentiment-analysis-on-sst-2-binary-classification | ERNIE | #38 | Accuracy: 93.5 |
| natural-language-inference-on-rte | ERNIE | #56 | Accuracy: 68.8% |