Natural Language Understanding on GLUE (SST-2, MNLI, MRPC, CoLA, QNLI, QQP, STS-B Subset)
88.92SST-2 AccuracyDWT
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DWTModel=RoBERTa-B2026.03 | 88.92 | 78.82 | 72.38 | 22.45 | 84.54 | 87.91 | 71.12 | 72.31 | |
| WSModel=RoBERTa-S2026.03 | 88.42 | 75.26 | 78.43 | 25.04 | 83.53 | 88.12 | 80.84 | 74.23 | |
| DWTModel=RoBERTa-S2026.03 | 88.42 | 76.08 | 78.43 | 23.27 | 85.58 | 88.62 | 82.81 | 74.74 | |
| ScratchModel=RoBERTa-S2026.03 | 88.19 | 75.3 | 76.47 | 21.27 | 85.19 | 88.48 | 82.87 | 73.97 | |
| LiGOModel=RoBERTa-B2026.03 | 87.61 | 76.87 | 71.08 | 21.72 | 83.4 | 87.83 | 70.86 | 71.34 | |
| MangoModel=RoBERTa-B2026.03 | 87.5 | 77.04 | 70.59 | 21.72 | 83.34 | 87.79 | 69.46 | 71.06 | |
| ScratchModel=RoBERTa-B2026.03 | 81.65 | 31.82 | 69.61 | 2.16 | 61.58 | 77.5 | 15.69 | 48.57 | |
| bert2BERTModel=RoBERTa-B2026.03 | 81.19 | 61.91 | 70.59 | 10.51 | 59.86 | 79.94 | 20.86 | 54.98 |