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

Misalignment Monitoring Dataset (dev)

0.962AUROC (Deception)

LLM Judge (Haiku)

0.828880.863440.8980.93256Jun 23, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
0.9620.9980.9640.970.9570.95986.7100
2026.06
0.9260.9710.9730.9880.940.97177.893
2026.06
0.910.9980.9670.9550.9610.9583.347
2026.06
0.890.8790.8950.9590.7650.90277.873
2026.06
0.8790.9970.9660.9880.9770.97688.933
2026.06
0.8340.9360.9510.8540.810.86555.6100