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

Safety and Helpfulness Evaluation on StrongREJECT

0.2Harm Rate

PPO SafeSearch

-2.60816.34635.354.254Oct 19, 2025
Updated 26d ago

Evaluation Results

MethodLinks
2025.10
0.22.65
2025.10
0.21.61
2025.10
0.52.59
2025.10
1.61.7
2025.10
1.93.32
2025.10
5.82.36
2025.10
7.91.84
2025.10
8.81.78
2025.10
11.23.15
2025.10
11.43.26
2025.10
14.32.15
2025.10
152.22
2025.10
17.31.74
2025.10
201.96
2025.10
27.32.19
2025.10
27.32.19
2025.10
31.12.18
2025.10
31.32.17
2025.10
35.22.27
2025.10
36.32.24
2025.10
36.32.24
2025.10
512.17
2025.10
52.72.33
2025.10
52.72.33
2025.10
55.32.2
2025.10
62.12.21
2025.10
63.52.22
2025.10
63.52.22
2025.10
70.42.12