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

Harmfulness Evaluation on DirectHarm (test)

5Harmfulness Score (Llama-Guard-3B)

Original

4.06810.35916.6522.941Mar 21, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.03
54.5
2025.03
54.5
2025.03
5.85.5
2025.03
7.57.3
2025.03
7.57.2
2025.03
7.57.1
2025.03
8.17.8
2025.03
8.17.8
2025.03
8.27.8
2025.03
8.57.9
2025.03
8.88.3
2025.03
9.18.8
2025.03
9.58.9
2025.03
10.19.6
2025.03
10.29.6
2025.03
10.39.8
2025.03
10.710.3
2025.03
11.110.6
2025.03
11.310.7
2025.03
11.310.7
2025.03
11.811.3
2025.03
11.911.6
2025.03
12.211.8
2025.03
12.511.9
2025.03
12.511.9
2025.03
12.512.1
2025.03
12.511.8
2025.03
13.112.6
2025.03
13.512.9
2025.03
13.713.4
2025.03
14.213.8
2025.03
14.213.7
2025.03
14.213.6
2025.03
14.914.1
2025.03
15.114.5
2025.03
15.714.4
2025.03
16.716.4
2025.03
17.116.4
2025.03
17.417.1
2025.03
17.616.9
2025.03
17.717.1
2025.03
18.217.6
2025.03
18.217.6
2025.03
18.317.6
2025.03
18.518
2025.03
18.818.4
2025.03
19.418.6
2025.03
19.519.1
2025.03
20.219.4
2025.03
22.121.4
2025.03
22.321.7
2025.03
23.523.1
2025.03
25.324.8
2025.03
2625.7
2025.03
27.827.3
2025.03
28.327.6