Malicious Package Detection on PyPI Malicious Package Dataset All Packages Combined (test)
97.08AccuracyH2GLM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| H2GLM2026.07 | 97.08 | 95.59 | 98.71 | 97.12 | |
| HERCULE2026.07 | 89.06 | 87.6 | 91 | 89.27 | |
| MPHunter2026.07 | 88.54 | 83.56 | 90.18 | 86.74 | |
| Qwen3-14bReasoning-enhanced=true2026.07 | 85.68 | 97.85 | 72.96 | 83.59 | |
| Warehouse2026.07 | 71.16 | 92.55 | 46.02 | 61.47 | |
| DeepSeek-R1-70bReasoning-enhanced=true2026.07 | 70.13 | 71.24 | 67.52 | 69.33 | |
| OSSGadget2026.07 | 63.82 | 53.11 | 91.57 | 67.23 | |
| Naive Bayes2026.07 | 53.42 | 53.21 | 55.62 | 54.39 | |
| Qwen2.5-7bReasoning-enhanced=true2026.07 | 45.4 | 44.39 | 36.42 | 40.01 | |
| Bandit4Mal2026.07 | 44.37 | 39.06 | 21.98 | 28.13 | |
| Llama3-8bReasoning-enhanced=true2026.07 | 43.72 | 45.61 | 65.2 | 53.67 | |
| Clustering2026.07 | 36.88 | 22.51 | 10.74 | 14.54 | |
| H2GLM2026.07 | 0.9708 | 0.9559 | 0.9871 | 0.9712 | |
| HERCULE2026.07 | 0.8906 | 0.876 | 0.91 | 0.8927 | |
| MPHunter2026.07 | 0.8854 | 0.8356 | 0.9018 | 0.8674 | |
| Qwen3-14b2026.07 | 0.8568 | 0.9785 | 0.7296 | 0.8359 | |
| Warehouse2026.07 | 0.7116 | 0.9255 | 0.4602 | 0.6147 | |
| DeepSeek-R1-70b2026.07 | 0.7013 | 0.7124 | 0.6752 | 0.6933 | |
| OSSGadget2026.07 | 0.6382 | 0.5311 | 0.9157 | 0.6723 | |
| Naive Bayes2026.07 | 0.5342 | 0.5321 | 0.5562 | 0.5439 | |
| Qwen2.5-7b2026.07 | 0.454 | 0.4439 | 0.3642 | 0.4001 | |
| Bandit4Mal2026.07 | 0.4437 | 0.3906 | 0.2198 | 0.2813 | |
| Llama3-8b2026.07 | 0.4372 | 0.4561 | 0.652 | 0.5367 | |
| Clustering2026.07 | 0.3688 | 0.2251 | 0.1074 | 0.1454 |