Malicious Package Detection on PyPI Minimal Imports (test)
96.6AccuracyH2GLM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| H2GLM2026.07 | 96.6 | 95.28 | 98.06 | 96.65 | |
| MPHunter2026.07 | 92.23 | 93.94 | 90.29 | 92.08 | |
| Qwen3-14bParameters=14b2026.07 | 88.83 | 98.78 | 78.64 | 87.57 | |
| OSSGadget2026.07 | 69.9 | 67.23 | 77.67 | 72.07 | |
| DeepSeek-R1-70bParameters=70b2026.07 | 65.05 | 64.76 | 66.02 | 65.38 | |
| HERCULE2026.07 | 63.11 | 68 | 49.51 | 57.3 | |
| Warehouse2026.07 | 59.71 | 100 | 19.42 | 32.52 | |
| Naive Bayes2026.07 | 52.13 | 62.5 | 10.64 | 18.18 | |
| Llama3-8bParameters=8b2026.07 | 49.51 | 49.64 | 66.99 | 57.02 | |
| Bandit4Mal2026.07 | 45.15 | 44.44 | 38.83 | 41.45 | |
| Qwen2.5-7bParameters=7b2026.07 | 43.69 | 43.3 | 40.78 | 42 | |
| Clustering2026.07 | 39.53 | 23.53 | 9.3 | 13.33 |