Legal Reasoning on LegalBench (MAE)
0.033MAEEnsemble
Evaluation Results
| Method | Links | |
|---|---|---|
| EnsembleNum. queries=64, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.033 | |
| DKPSNum. queries=64, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.034 | |
| Sample ScoreNum. queries=64, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.048 | |
| EnsembleNum. queries=16, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.049 | |
| DKPSNum. queries=16, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.051 | |
| DKPSNum. queries=4, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.067 | |
| EnsembleNum. queries=4, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.067 | |
| DKPSNum. queries=1, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.087 | |
| EnsembleNum. queries=1, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.087 | |
| Pop. MeanNum. queries=1, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.093 | |
| Pop. MeanNum. queries=4, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.093 | |
| Pop. MeanNum. queries=16, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.093 | |
| Pop. MeanNum. queries=64, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.093 | |
| Sample ScoreNum. queries=16, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.097 | |
| Sample ScoreNum. queries=4, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.199 | |
| Sample ScoreNum. queries=1, Evaluation protocol=Leave-One-Family-Out protocol2026.05 | 0.472 |