NLU, QA, and NER on GLUE, SuperGLUE, MRQA, and WNUT2017 NER v1 (dev)
89.4GLUE ScoreCA-MTL
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CA-MTLBackbone=ROBERTa-LARGE, Total Params=1.12x2020.09 | 89.4 | 80 | 82.4 | 55.2 | 83.1 | — | |
| STReImpBackbone=ROBERTa-LARGE, Total Params=24x2020.09 | 88.2 | 76.5 | 83.6 | 57.8 | 81.9 | — | |
| CA-MTLBackbone=BERT-LARGE, Total Params=1.12x2020.09 | 86.6 | 74.1 | 79.5 | 49 | 79.1 | — | |
| MTLReImpBackbone=ROBERTa-LARGE, Total Params=1x2020.09 | 86 | 78.6 | 80.7 | 49.3 | 80.7 | — | |
| STReImpBackbone=BERT-LARGE, Total Params=24x2020.09 | 84.5 | 68.9 | 79.7 | 54.1 | 76.8 | — | |
| MTLReImpBackbone=BERT-LARGE, Total Params=1x2020.09 | 83.2 | 72.1 | 77.8 | 42.2 | 76.4 | — |