Malicious Prompt Detection on JailbreakV_28K Text-based (test)
0FNRVLMShield
Evaluation Results
| Method | Links | |
|---|---|---|
| VLMShieldTarget VLM Architecture=LLaVA-1.5-13B2026.04 | 0 | |
| VLMShieldTarget VLM Architecture=Qwen2.5-VL-7B-Instruct2026.04 | 0 | |
| ASTRATarget VLM Architecture=Qwen2.5-VL-7B-Instruct2026.04 | 1.72 | |
| ASTRATarget VLM Architecture=LLaVA-1.5-13B2026.04 | 3.88 | |
| VLMGuardTarget VLM Architecture=Qwen2.5-VL-7B-Instruct2026.04 | 5.72 | |
| SelfReminderTarget VLM Architecture=Qwen2.5-VL-7B-Instruct2026.04 | 8.4 | |
| VLMGuardTarget VLM Architecture=LLaVA-1.5-13B2026.04 | 9.26 | |
| JailGuardTarget VLM Architecture=Qwen2.5-VL-7B-Instruct2026.04 | 16.18 | |
| MirrorCheckTarget VLM Architecture=LLaVA-1.5-13B2026.04 | 20.65 | |
| MirrorCheckTarget VLM Architecture=Qwen2.5-VL-7B-Instruct2026.04 | 20.65 | |
| ECSOTarget VLM Architecture=Qwen2.5-VL-7B-Instruct2026.04 | 22.83 | |
| JailGuardTarget VLM Architecture=LLaVA-1.5-13B2026.04 | 26.33 | |
| ECSOTarget VLM Architecture=LLaVA-1.5-13B2026.04 | 28.06 | |
| CIDERTarget VLM Architecture=LLaVA-1.5-13B2026.04 | 48.53 | |
| CIDERTarget VLM Architecture=Qwen2.5-VL-7B-Instruct2026.04 | 48.53 | |
| SelfReminderTarget VLM Architecture=LLaVA-1.5-13B2026.04 | 70.87 |