Predictive Validity Verification on Downstream Tasks (KOLD, HateXplain, etc.)
0.3156Average CorrelationDOVE
Evaluation Results
| Method | Links | |
|---|---|---|
| DOVEEvaluation Mode=Average across 12 LLMs2026.03 | 0.3156 | |
| CDEvalEvaluation Mode=Average across 12 LLMs2026.03 | 0.2356 | |
| WVSEvaluation Mode=Average across 12 LLMs2026.03 | 0.162 | |
| NormAdEvaluation Mode=Average across 12 LLMs2026.03 | 0.009 | |
| GOQAEvaluation Mode=Average across 12 LLMs2026.03 | -0.1305 | |
| NaVABEvaluation Mode=Average across 12 LLMs2026.03 | -0.2077 |