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

Safety Alignment on Do-Not-Answer

0MD

ShaPO-R

-2.490414.319831.1347.9402Feb 7, 2026Feb 11, 2026Feb 16, 2026Feb 21, 2026Feb 25, 2026Mar 2, 2026Mar 7, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.02
00
2026.02
0.110.11
2026.02
0.110
2026.02
0.210
2026.03
0.210
2026.03
0.210
2026.02
0.430.11
2026.02
0.530
2026.02
0.750
2026.02
0.960
2026.03
0.960
2026.02
1.070
2026.03
1.070.11
2026.02
1.170.64
2026.02
1.170.53
2026.02
1.170.32
2026.03
1.170.53
2026.03
1.170.32
2026.02
3.731.71
2026.02
5.010.64
2026.02
5.012.03
2026.02
6.720.43
2026.02
10.343.41
2026.03
10.343.41
2026.02
10.984.16
2026.03
10.984.16
2026.02
11.515.54
2026.02
14.187.78
2026.03
14.187.78
2026.02
14.297.36
2026.03
14.297.36
2026.02
16.429.81
2026.02
16.425.76
2026.03
16.425.76
2026.02
21.117.78
2026.03
21.117.78
2026.02
22.3913.86
2026.03
22.3913.86
2026.02
39.2314.29
2026.02
44.9916.52
2026.02
46.2614.5
2026.02
47.5513.33
2026.02
51.7127.72
2026.02
52.4511.3
2026.02
52.6731.56
2026.03
52.6731.56
2026.02
53.5233.48
2026.03
53.5233.48
2026.02
54.5812.37
2026.03
54.5812.37
2026.02
59.718.34
2026.02
62.2618.44