Model Performance: | |
precision recall f1-score support | |
Normal (0) 1.00 0.90 0.95 39 | |
Malicious (1) 0.91 1.00 0.96 43 | |
accuracy 0.95 82 | |
macro avg 0.96 0.95 0.95 82 | |
weighted avg 0.96 0.95 0.95 82 | |
Confusion Matrix: | |
[[35 4] | |
[ 0 43]] | |
Notes: | |
- Precision: Ability to avoid false positives | |
- Recall: Ability to detect all malicious prompts | |
- F1-score: Balanced measure of precision and recall | |