Text-based Jailbreak on JailbreakV_28K IND
70.87Attack Success Rate (ASR)SelfReminder
Evaluation Results
| Method | Links | |
|---|---|---|
| SelfReminderBase Model=LLaVA-1.5-13B2026.04 | 70.87 | |
| CIDERBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.04 | 48.53 | |
| CIDERBase Model=LLaVA-1.5-13B2026.04 | 48.53 | |
| ECSOBase Model=LLaVA-1.5-13B2026.04 | 28.06 | |
| JailGuardBase Model=LLaVA-1.5-13B2026.04 | 26.33 | |
| ECSOBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.04 | 22.83 | |
| MirrorCheckBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.04 | 20.65 | |
| MirrorCheckBase Model=LLaVA-1.5-13B2026.04 | 20.65 | |
| JailGuardBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.04 | 16.18 | |
| VLMGuardBase Model=LLaVA-1.5-13B2026.04 | 9.26 | |
| SelfReminderBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.04 | 8.4 | |
| VLMGuardBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=Internal Defense2026.04 | 5.72 | |
| ASTRABase Model=LLaVA-1.5-13B2026.04 | 3.88 | |
| ASTRABase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=Internal Defense2026.04 | 1.72 | |
| VLMShieldBase VLM=Qwen2.5-VL-7B-Instruct, Defense Category=External Defense2026.04 | 0 | |
| VLMShieldBase Model=LLaVA-1.5-13B2026.04 | 0 |