List your company

Prompt Injection Attack Detection on XSum

100TPRDualSentinel
8790.37593.7597.125Mar 2, 2026

Evaluation results

MethodLinks
DualSentinelBackbone=GPT-4o, Attacks=System, Target Form=OnlyTarget, Sequence Length=Long2026.031000
DualSentinelBackbone=GPT-4o-mini, Attacks=System, Target Form=OnlyTarget, Sequence Length=Long2026.0310040
DualSentinelModel=LlaMA2 13B, Attacks=Tree, Target format=OnlyTarget, Target sequence length=Long2026.031002.8
DualSentinelModel=LlaMA2 7B, Attacks=Word, Target format=OnlyTarget, Target sequence length=Long2026.031008
DualSentinelModel=LlaMA2 7B, Attacks=Phrase, Target format=OnlyTarget, Target sequence length=Long2026.031007
DualSentinelModel=Qwen2.5 14B, Attacks=Complete, Target format=OnlyTarget, Target sequence length=Long2026.031001.6
DualSentinelModel=Qwen2.5 14B, Attacks=System, Target format=OnlyTarget, Target sequence length=Long2026.031001.6
DualSentinelModel=Qwen2.5 7B, Attacks=Word, Target format=OnlyTarget, Target sequence length=Long2026.031004.82
DualSentinelModel=Qwen2.5 7B, Attacks=Joint, Target format=OnlyTarget, Target sequence length=Long2026.031005.71
DualSentinelBackbone=GPT-4o, Attacks=Average, Target Form=OnlyTarget, Sequence Length=Long2026.0399.80
DualSentinelModel=Qwen2.5 14B, Attacks=Ignore, Target format=OnlyTarget, Target sequence length=Long2026.0399.691.6
DualSentinelBackbone=GPT-4o, Attacks=Complete, Target Form=OnlyTarget, Sequence Length=Long2026.0399.60
DualSentinelModel=LlaMA2 7B, Attacks=Sentence, Target format=OnlyTarget, Target sequence length=Long2026.039910
DualSentinelModel=LlaMA2 7B, Attacks=Joint, Target format=OnlyTarget, Target sequence length=Long2026.03998
DualSentinelModel=Qwen2.5 7B, Attacks=Phrase, Target format=OnlyTarget, Target sequence length=Long2026.03997.14
DualSentinelModel=Qwen2.5 7B, Attacks=Sentence, Target format=OnlyTarget, Target sequence length=Long2026.03997.77
DualSentinelBackbone=GPT-4o-mini, Attacks=Average, Target Form=OnlyTarget, Sequence Length=Long2026.0398.640
DualSentinelModel=LlaMA2 13B, Attacks=System, Target format=OnlyTarget, Target sequence length=Long2026.0398.32.8
DualSentinelModel=Qwen2.5 14B, Attacks=Tree, Target format=OnlyTarget, Target sequence length=Long2026.0397.81.6
DualSentinelBackbone=GPT-4o-mini, Attacks=Complete, Target Form=OnlyTarget, Sequence Length=Long2026.0397.240
DualSentinelModel=LlaMA2 13B, Attacks=Ignore, Target format=OnlyTarget, Target sequence length=Long2026.0388.552.8
DualSentinelModel=LlaMA2 13B, Attacks=Complete, Target format=OnlyTarget, Target sequence length=Long2026.0387.52.8