Research Article
VulDistilBERT: A CPS Vulnerability Severity Prediction Method Based on Distillation Model
Table 7
Results of evaluating the raw dataset with the fine-tuned model.
| Models | Best epoch | Loss | Accuracy | Precision | Recall | F1 scores |
| Fine-tuned BERT | 2 | 0.580 | 0.77 | 0.75 | 0.62 | 0.66 | Fine-tuned XLNET | 4 | 0.570 | 0.78 | 0.74 | 0.65 | 0.68 | Fine-tuned ROBERTA | 4 | 0.566 | 0.78 | 0.76 | 0.65 | 0.68 | Fine-tuned DistilBERT | 4 | 0.571 | 0.78 | 0.76 | 0.68 | 0.70 |
|
|