WMDP
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
WMDP chem
1.351CAD Score
16
WMDP bio
0CAD
16
WMDP
76.1Unlearn Score
16
WMDP
100Accuracy
16
WMDP cyber
41BF16 Precision Score
15
WMDP chem
42.3BF16 Score
15
WMDP bio
55.7BF16 Score
15
WMDP Bio
75.1MMLU
15
WMDP
64.7Bio Score
15
WMDP v1 (test)
42.43Accuracy (Cyber)
14
WMDP
62.4Delta Score
14
WMDP
50.95Overall Score
13
WMDP cyber
44.5Unlearning Accuracy
13
WMDP
79.6Bio Accuracy
12
WMDP
9Alignment Faking Rate
11
WMDP Bio (test)
64.81Accuracy Forget
11
WMDP Bio
78.5Acc_r
10
WMDP (test)
100TSR
10
WMDP average of biology and cyber
0.557Accuracy
10
WMDP Bio (test)
63.7Bio Score
10
WMDP bio
0.628WMDPbio Score
9
WMDP
0.489WMDP Score
9
WMDP bio
3.1Multi-turn ASR Error Rate
9
WMDP bio
48.9Accuracy
9
WMDP
39.7Accuracy
9