# Refusal Rate Evaluation on VQA v2 Source: [Refusal Rate Evaluation on VQA v2]() Collection index: [llms.txt]() Data updated: 2026-04-13T08:39:55.257Z Evaluation of model refusal behavior on the VQA v2 dataset, measured by the Refusal Rate. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Refusal Rate Evaluation | | Dataset (including any stated split) | VQA v2 | | Metric | Refusal Rate | | Unit | PERCENT | | Direction | Lower is better | | Results loaded | 30 of 30 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Refusal Rate](): Lower is better; unit: PERCENT. ## Recorded results | Method | Refusal Rate | Result date | Report paper | | --- | --- | --- | --- | | ImageProtector | 0.77 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.8 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.83 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.85 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.86 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.87 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.89 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.9 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.9 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.91 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.91 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.91 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.92 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.93 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.93 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.94 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.94 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.94 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.94 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.94 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.95 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.96 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 0.99 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 1 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 88 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 88 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 89 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 93 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 94 | Not recorded | 69dc78627d0172140ec7891f | | ImageProtector | 94 | Not recorded | 69dc78627d0172140ec7891f | ### ImageProtector Refusal Rate: 0.77. Result ID: 69dcaac147f34ce924c3a16c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"InstructBLIP","Shadow Question Type":"general probing questions","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.8. Result ID: 69dcaac147f34ce924c3a16e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Qwen2.5-VL","Shadow Question Type":"general probing questions","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.83. Result ID: 69dcaac147f34ce924c3a1b9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"InstructBLIP","Shadow Question Type":"similar","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.85. Result ID: 69dcaac147f34ce924c3a16d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Phi-4-multimodal","Shadow Question Type":"general probing questions","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.86. Result ID: 69dcaac147f34ce924c3a19f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"MiniGPT-4","Shadow Question Type":"exact","Question Relevance":"image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.87. Result ID: 69dcaac147f34ce924c3a16b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Qwen-VL-Chat","Shadow Question Type":"general probing questions","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.89. Result ID: 69dcaac147f34ce924c3a189. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"InstructBLIP","Probing Question Type":"exact","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.9. Result ID: 69dcaac147f34ce924c3a169. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"LLaVA-1.5","Shadow Question Type":"general probing questions","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.9. Result ID: 69dcaac147f34ce924c3a187. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"MiniGPT-4","Probing Question Type":"exact","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.91. Result ID: 69dcaac147f34ce924c3a186. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"LLaVA-1.5","Probing Question Type":"exact","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.91. Result ID: 69dcaac147f34ce924c3a1a1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"InstructBLIP","Shadow Question Type":"exact","Question Relevance":"image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.91. Result ID: 69dcaac147f34ce924c3a1b8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Qwen-VL-Chat","Shadow Question Type":"similar","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.92. Result ID: 69dcaac147f34ce924c3a1b6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"LLaVA-1.5","Shadow Question Type":"similar","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.93. Result ID: 69dcaac147f34ce924c3a188. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Qwen-VL-Chat","Probing Question Type":"exact","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.93. Result ID: 69dcaac147f34ce924c3a1b7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"MiniGPT-4","Shadow Question Type":"similar","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.94. Result ID: 69dcaac147f34ce924c3a16a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"MiniGPT-4","Shadow Question Type":"general probing questions","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.94. Result ID: 69dcaac147f34ce924c3a19e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"LLaVA-1.5","Shadow Question Type":"exact","Question Relevance":"image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.94. Result ID: 69dcaac147f34ce924c3a1a0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Qwen-VL-Chat","Shadow Question Type":"exact","Question Relevance":"image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.94. Result ID: 69dcaac147f34ce924c3a1ba. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Phi-4-multimodal","Shadow Question Type":"similar","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.94. Result ID: 69dcaac147f34ce924c3a1bb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Qwen2.5-VL","Shadow Question Type":"similar","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.95. Result ID: 69dcaac147f34ce924c3a18b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Qwen2.5-VL","Probing Question Type":"exact","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.96. Result ID: 69dcaac147f34ce924c3a1a3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Qwen2.5-VL","Shadow Question Type":"exact","Question Relevance":"image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 0.99. Result ID: 69dcaac147f34ce924c3a18a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Phi-4-multimodal","Probing Question Type":"exact","Question Relevance":"image-irrelevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 1. Result ID: 69dcaac147f34ce924c3a1a2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Phi-4-multimodal","Shadow Question Type":"exact","Question Relevance":"image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 88. Result ID: 69dcaac047f34ce924c3a151. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"LLaVA-1.5","Shadow Question Type":"Similar probing questions","Question Relevance":"Image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 88. Result ID: 69dcaac047f34ce924c3a152. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"MiniGPT-4","Shadow Question Type":"Similar probing questions","Question Relevance":"Image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 89. Result ID: 69dcaac047f34ce924c3a154. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"InstructBLIP","Shadow Question Type":"Similar probing questions","Question Relevance":"Image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 93. Result ID: 69dcaac047f34ce924c3a155. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Phi-4-multimodal","Shadow Question Type":"Similar probing questions","Question Relevance":"Image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 94. Result ID: 69dcaac047f34ce924c3a153. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Qwen-VL-Chat","Shadow Question Type":"Similar probing questions","Question Relevance":"Image-relevant"}. This record does not assert independent reproduction. ### ImageProtector Refusal Rate: 94. Result ID: 69dcaac047f34ce924c3a156. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Target MLLM":"Qwen2.5-VL","Shadow Question Type":"Similar probing questions","Question Relevance":"Image-relevant"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.