Out-of-distribution Question Answering on CPIQA Adversarial Attack
84.14AccuracyML Routing
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ML RoutingEvaluation Context=non-refused test data, Domain=Out-of-distribution (OOD), Attack=TextAttack2026.06 | 84.14 | 84.14 | 79.11 | 25.89 | |
| LWV FusionEvaluation Context=non-refused test data, Domain=Out-of-distribution (OOD), Attack=TextAttack2026.06 | 80.43 | 80.43 | 58.8 | 42.12 | |
| FreeLBEvaluation Context=non-refused test data, Domain=Out-of-distribution (OOD), Attack=TextAttack2026.06 | 34.2 | 40.27 | — | 70.94 | |
| Base ModelEvaluation Context=non-refused test data, Domain=Out-of-distribution (OOD), Attack=TextAttack2026.06 | 27 | 27 | — | 78.8 |