Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Embedded Direct Attack on Samsum ⊕ HEx-PHI
Loading...
52.4
Utility
Lisa
47.72
48.935
50.15
51.365
Jun 29, 2026
Utility
ASR
Updated 25d ago
Evaluation Results
Method
Method
Links
Utility
ASR
Lisa
Model=Qwen2.5-7B-Instruct
2026.06
52.4
62.7
Constrained SFT
Model=Qwen2.5-7B-Instruct
2026.06
52.3
74.1
SFT
Model=Llama-2-7B-Chat
2026.06
51.5
77.2
Lisa
Model=Llama-2-7B-Chat
2026.06
51.4
65.5
SFT
Model=Qwen2.5-7B-Instruct
2026.06
51.1
83.3
Lisa
Model=Gemma-3-4B-IT
2026.06
51
82.4
Constrained SFT
Model=Gemma-3-4B-IT
2026.06
50.7
82.7
SFT
Model=Gemma-3-4B-IT
2026.06
50.5
90
Constrained SFT
Model=Llama-2-7B-Chat
2026.06
50.4
74.7
DR-SFT
Model=Qwen2.5-7B-Instruct
2026.06
50.1
40.9
DR-SFT
Model=Llama-2-7B-Chat
2026.06
49.1
48.7
DR-SFT
Model=Gemma-3-4B-IT
2026.06
47.9
75.4
Feedback
Search any
task
Search any
task