List your company

Invocation attack on ToolBench

98CDAFlip-Agent
68.8876.448491.56Mar 7, 2026

Evaluation results

MethodLinks
Flip-AgentModel=Llama-3.2-3B-Instruct2026.039898.9
Flip-AgentModel=ToolLLaMA-2-7b-v22026.0398100
Flip-SModel=ToolLLaMA-2-7b-v22026.0397100
Flip-AgentModel=Qwen3-8B2026.039767.3
Flip-SModel=Qwen3-4B-Instruct-25072026.039676.8
TBTModel=ToolLLaMA-2-7b-v22026.03950
TrojViTModel=ToolLLaMA-2-7b-v22026.039540
Flip-AgentModel=Llama-3.1-8B-Instruct2026.039597.2
TBTModel=Qwen3-8B2026.03950
Flip-AgentModel=Qwen3-4B-Instruct-25072026.039587.3
TrojViTModel=Llama-3.1-8B-Instruct2026.039265
TrojViTModel=Llama-3.2-3B-Instruct2026.039055
Flip-SModel=Llama-3.1-8B-Instruct2026.039096.8
TrojViTModel=Qwen3-8B2026.039060
TBTModel=DeepSeek-R1-Distill-Llama-8B2026.03905
TrojViTModel=DeepSeek-R1-Distill-Llama-8B2026.039060
Flip-AgentModel=DeepSeek-R1-Distill-Llama-8B2026.039082.1
Flip-SModel=Llama-3.2-3B-Instruct2026.038583.1
TrojViTModel=Qwen3-4B-Instruct-25072026.038560
Flip-SModel=DeepSeek-R1-Distill-Llama-8B2026.038576.8
Flip-SModel=Qwen3-8B2026.038163.1
TBTModel=Llama-3.2-3B-Instruct2026.03780
TBTModel=Llama-3.1-8B-Instruct2026.037025
TBTModel=Qwen3-4B-Instruct-25072026.037010