Code Generation on LCC
50ASRNo Defense
Evaluation results
| Method | Links | ||
|---|---|---|---|
| No DefenseAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 50 | — | |
| PromptArmorAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 48 | — | |
| No DefenseAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 32 | — | |
| PIGuardAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 32 | — | |
| PromptArmorAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 30 | — | |
| PromptGuardAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 26 | — | |
| PromptGuardAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 20 | — | |
| PIGuardAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 20 | — | |
| GPT-OSS-SafeguardAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 14 | — | |
| DataSentinelAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 10 | — | |
| AgentWatcherAttack Type=Direct, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 3 | — | |
| GPT-OSS-SafeguardAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 2 | — | |
| AgentWatcherAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 2 | — | |
| DataSentinelAttack Type=Combined, Target LLM=Qwen-3-4B-Instruct-25072026.04 | 1 | — | |
| AgentWatcherAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 67 | |
| DataSentinelAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 40 | |
| GPT-OSS-SafeguardAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 67 | |
| No DefenseAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 67 | |
| PIGuardAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 58 | |
| PromptArmorAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 67 | |
| PromptGuardAttack Type=Clean, Target LLM=Qwen-3-4B-Instruct-25072026.04 | — | 61 |