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

Safety Evaluation on Do-Not-Answer

62.26MD Rate

Pythia-2.8B + SFT

-1.824814.812631.4548.0874Mar 7, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.03
62.2618.44
2026.03
59.718.34
2026.03
52.4511.3
2026.03
51.7127.72
2026.03
47.5513.33
2026.03
46.2614.5
2026.03
44.9916.52
2026.03
39.2314.29
2026.03
11.515.54
2026.03
8.961.28
2026.03
6.821.07
2026.03
5.012.03
2026.03
3.731.71
2026.03
1.170.64
2026.03
1.070
2026.03
0.640