Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Backdoor Detection on Simple IHU Llama 8B

0.992AUROC

Functional Attribution

0.433520.578510.72350.86849Apr 21, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
0.99210099
2026.04
0.98810088
2026.04
0.9879848
2026.04
0.97610099
2026.04
0.97410099
2026.04
0.97310099
2026.04
0.94810088
2026.04
0.94310088
2026.04
0.83210088
2026.04
0.8310099
2026.04
0.8259848
2026.04
0.799848
2026.04
0.73110088
2026.04
0.6829848
2026.04
0.4559848