Malicious Package Detection on PyPI (test)
96.1PrecisionCerebro RoBERTa
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Cerebro RoBERTaTrain=Mixed2023.09 | 96.1 | 90.9 | |
| Cerebro RoBERTaTrain=PyPI, Backbone=RoBERTa, Learning-based=true2023.09 | 96 | 91.7 | |
| Cerebro BERTTrain=PyPI, Backbone=BERT, Learning-based=true2023.09 | 95.8 | 89.4 | |
| Cerebro BERTTrain=Mixed2023.09 | 95 | 92 | |
| Cerebro T5Train=Mixed2023.09 | 94.6 | 69.6 | |
| Cerebro T5Train=PyPI, Backbone=T5, Learning-based=true2023.09 | 93.4 | 70.9 | |
| SAPTrain=PyPI, Learning-based=true2023.09 | 83.9 | 61.5 | |
| AmalfiDTTrain=Mixed2023.09 | 83.2 | 81.1 | |
| AmalfiDTTrain=PyPI, Algorithm=Decision Tree, Learning-based=true2023.09 | 81.9 | 83.9 | |
| SAPTrain=Mixed2023.09 | 80.2 | 60.2 | |
| AmalfiNBTrain=PyPI, Algorithm=Naive Bayes, Learning-based=true2023.09 | 59.5 | 7.5 | |
| AmalfiSVMTrain=Mixed2023.09 | 45.1 | 37.2 | |
| AmalfiNBTrain=Mixed2023.09 | 42 | 35.6 | |
| AmalfiSVMTrain=PyPI, Algorithm=SVM, Learning-based=true2023.09 | 30.3 | 30.9 | |
| Bandit4MalApproach Type=Rule-Based2023.09 | 24.9 | 74.4 | |
| MPHunterTrain=PyPI, Learning-based=true2023.09 | 21.8 | 85.4 | |
| OSS Detect BackdoorApproach Type=Rule-Based2023.09 | 19.9 | 60.4 |