Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Prompt Injection Defense on GSM8K PI (test)

3.3ASR@1

No Defense

3.0045.00278.998Jan 21, 2026
Updated 4d ago

Evaluation Results

MethodLinks
2026.01
3.37.71196.793.390
2026.01
3.38.710.796.791.790
2026.01
3.33.73.396.796.796.7
2026.01
3.73.33.396.796.796.7
2026.01
3.77.310.396.793.390
2026.01
4.35.77.3959593.3
2026.01
5.514.515.99384.784.7
2026.01
6.7131893.386.781.7
2026.01
6.76.76.793.393.393.3
2026.01
6.76.76.793.393.393.3
2026.01
721.731.393.38071.7
2026.01
7.210.316.192.392.382.1
2026.01
911.714.288.376.771.7
2026.01
9.32026.79083.375
2026.01
10.312.212.488.38583.3
2026.01
10.726.335.788.373.365