Text Classification on Reuters BERT-base (test)
95.7AccuracyAdapterFusion
Evaluation results
| Method | Links | |
|---|---|---|
| AdapterFusionNum Params=182 M, Backbone=BERT-base2022.03 | 95.7 | |
| ADA (K=4) - TransRateNum Params=128 M, Backbone=BERT-base2022.03 | 95.1 | |
| ADA (K=4) - LEEPNum Params=128 M, Backbone=BERT-base2022.03 | 94.7 | |
| AdaptersNum Params=182 M, Backbone=BERT-base2022.03 | 91.8 | |
| ADA (K=1)Num Params=120.8 M, Backbone=BERT-base2022.03 | 82.5 |