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

Prompt Injection on CSQA

18.33ASR@3

G-Safeguard

17.183224.924132.66540.4059Dec 21, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.12
18.33100
2025.12
18.33100
2025.12
18.6798.22
2025.12
19.67100
2025.12
20.6791.11
2025.12
21.6787.11
2025.12
22.6689.78
2025.12
23.3383.56
2025.12
23.6777.78
2025.12
23.6755.56
2025.12
2554.22
2025.12
2575.11
2025.12
2575.11
2025.12
2590.67
2025.12
25.6746.67
2025.12
26.3344.89
2025.12
26.3352
2025.12
2854.67
2025.12
30.3350.67
2025.12
31.3353.78
2025.12
31.3344.44
2025.12
32.67-
2025.12
3445.33
2025.12
34.3356.44
2025.12
38-
2025.12
40.3344
2025.12
43-
2025.12
47-