Natural Language Inference on HANS (OOD)
65.82AccuracyQ-Diversity
Evaluation results
| Method | Links | |
|---|---|---|
| Q-DiversityTraining Source=MultiNLI2023.05 | 65.82 | |
| JTTTraining Source=MultiNLI2023.05 | 65.32 | |
| EIILTraining Source=MultiNLI2023.05 | 65.06 | |
| ERMTraining Source=MultiNLI2023.05 | 62.11 |