Active Learning on Civil Comments (test)
89.57AccuracyParBaLS EPIG
Evaluation results
| Method | Links | |
|---|---|---|
| ParBaLS EPIGBudget=100+10x20, Encoder=bert_base_uncased, PCA Dimensions=1872025.10 | 89.57 | |
| ParBaLS-MAP EPIGBudget=100+10x20, Encoder=bert_base_uncased, PCA Dimensions=1872025.10 | 88.8 | |
| EPIGBudget=100+10x20, Encoder=bert_base_uncased, PCA Dimensions=1872025.10 | 88.41 | |
| SoftRankEPIGBudget=100+10x20, Encoder=bert_base_uncased, PCA Dimensions=1872025.10 | 87.37 | |
| PowerBALDBudget=100+10x20, Encoder=bert_base_uncased, PCA Dimensions=1872025.10 | 86.67 | |
| SoftRankBALDBudget=100+10x20, Encoder=bert_base_uncased, PCA Dimensions=1872025.10 | 86.35 | |
| SoftmaxEPIGBudget=100+10x20, Encoder=bert_base_uncased, PCA Dimensions=1872025.10 | 85.31 | |
| SoftmaxBALDBudget=100+10x20, Encoder=bert_base_uncased, PCA Dimensions=1872025.10 | 85.25 | |
| PowerEPIGBudget=100+10x20, Encoder=bert_base_uncased, PCA Dimensions=1872025.10 | 83.92 | |
| BALDBudget=100+10x20, Encoder=bert_base_uncased, PCA Dimensions=1872025.10 | 82.19 |