Explanation Generation on Stanford Dogs
1EEL+GE
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| L+GEtraining_mode=train with labels and general explanations2025.02 | 1 | 0.77 | 7.5 | |
| SelfSynthXtraining_mode=rejection sampling with detailed visual knowledge2025.02 | 1 | 0.86 | 7.41 | |
| Basetraining_mode=original model2025.02 | 0.94 | 0.69 | 5.47 | |
| NLtraining_mode=only train with labels2025.02 | 0 | — | — |