Whataboutism Detection on Twitter replies (test)
0.91Precision (W)MINA
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MINABackbone=ROBERTA Irony, Params (M)=1332024.02 | 0.91 | 0.57 | 0.7 | 0.54 | 0.89 | 0.67 | |
| MINABackbone=SBERT, Params (M)=312024.02 | 0.73 | 0.78 | 0.75 | 0.87 | 0.84 | 0.85 | |
| BERTApproach=Transfer learning (baseline models), Params (M)=1102024.02 | 0.64 | 0.76 | 0.69 | 0.85 | 0.76 | 0.8 | |
| ROBERTAApproach=Transfer learning (baseline models), Params (M)=1242024.02 | 0.64 | 0.8 | 0.71 | 0.87 | 0.75 | 0.81 | |
| SBERTApproach=Semantic similarity, Params (M)=22.72024.02 | 0.64 | 0.76 | 0.69 | 0.85 | 0.76 | 0.8 | |
| SimCSEApproach=Semantic similarity, Params (M)=22.72024.02 | 0.64 | 0.8 | 0.71 | 0.87 | 0.74 | 0.8 | |
| C&SApproach=Semantic similarity, Params (M)=22.72024.02 | 0.57 | 0.6 | 0.59 | 0.95 | 0.91 | 0.93 | |
| ROBERTA IronyApproach=Transfer learning (baseline models), Params (M)=1242024.02 | 0.54 | 0.89 | 0.67 | 0.91 | 0.57 | 0.7 |