Explanation Generation on PLD
100EEL+GE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| L+GEtraining_mode=train with labels and general explanations2025.02 | 100 | 0.76 | 10.45 | |
| SelfSynthXtraining_mode=rejection sampling with detailed visual knowledge2025.02 | 100 | 0.86 | 9.01 | |
| Basetraining_mode=original model2025.02 | 95 | — | — | |
| NLtraining_mode=only train with labels2025.02 | 0 | — | — |