# Meme Abuse Detection on GOAT-Bench 1.0 (test) Source: [Meme Abuse Detection on GOAT-Bench 1.0 (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:46:23.561Z Evaluation of model performance on the test split of the GOAT-Bench 1.0 dataset for detecting various forms of meme abuse, including harmfulness, hatefulness, misogyny, offensiveness, and sarcasm. 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 | Meme Abuse Detection | | Dataset (including any stated split) | GOAT-Bench 1.0 (test) | | Metric | Harmfulness Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 32 of 32 | | 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 - [Harmfulness Accuracy](): Higher is better; unit: PERCENT. - [Harmfulness F1 Score](): Higher is better; unit: PERCENT. - [Hatefulness Accuracy](): Higher is better; unit: PERCENT. - [Hatefulness F1 Score](): Higher is better; unit: PERCENT. - [Misogyny Accuracy](): Higher is better; unit: PERCENT. - [Misogyny F1 Score](): Higher is better; unit: PERCENT. - [Offensiveness Accuracy](): Higher is better; unit: PERCENT. - [Offensiveness F1 Score](): Higher is better; unit: PERCENT. - [Sarcasm Accuracy](): Higher is better; unit: PERCENT. - [Sarcasm F1 Score](): Higher is better; unit: PERCENT. - [Overall Accuracy](): Higher is better; unit: PERCENT. - [Overall F1 Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Harmfulness Accuracy | Result date | Report paper | | --- | --- | --- | --- | | CROSS-ALIGN+ | 75.1 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 72.8 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 71 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 69.9 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 69.5 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 69.3 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 68.6 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 68.1 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 67.5 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 67 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 66.9 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 66.8 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 66.7 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 65.2 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 64.3 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | CROSS-ALIGN+ | 63.9 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | Qwen2.5-VL | 63.5 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | LLaVA-V1.6 | 61.7 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | InternVL2 | 59.8 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | DeepSeek-VL-S | 58.3 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | mPLUG-Owl2-Small | 58 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | Qwen2.5-VL | 57.8 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | MiniGPT-4 | 57 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | LLaVA-V1.6 | 56.9 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | Gemma-VL | 56.5 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | InternVL2 | 55.9 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | mPLUG-Owl2-Small | 55.6 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | DeepSeek-VL-S | 54.9 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | BLIP-2 OPT | 54.9 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | MiniGPT-4 | 53.8 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | Gemma-VL | 52.6 | Not recorded | 6982d2fea8b01e76cc8e9a51 | | BLIP-2 OPT | 52.3 | Not recorded | 6982d2fea8b01e76cc8e9a51 | ### CROSS-ALIGN+ Harmfulness Accuracy: 75.1. Result ID: 69964e415698665ec492d846. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL (7B)","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 72.8. Result ID: 69964e415698665ec492d844. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-V1.6 (7B)","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 71. Result ID: 69964e415698665ec492d848. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InternVL2 (7B)","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 69.9. Result ID: 69964e415698665ec492d84c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"DeepSeek-VL-S (7B)","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 69.5. Result ID: 69964e415698665ec492d836. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL (7B)","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 69.3. Result ID: 69964e415698665ec492d850. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"mPLUG-Owl2-Small (7B)","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 68.6. Result ID: 69964e415698665ec492d84e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"MiniGPT-4 (7B)","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 68.1. Result ID: 69964e415698665ec492d84a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Gemma-VL (7B)","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 67.5. Result ID: 69964e415698665ec492d834. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"LLaVA-V1.6 (7B)","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 67. Result ID: 69964e415698665ec492d838. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"InternVL2 (7B)","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 66.9. Result ID: 69964e415698665ec492d840. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"mPLUG-Owl2-Small (7B)","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 66.8. Result ID: 69964e415698665ec492d83c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"DeepSeek-VL-S (7B)","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 66.7. Result ID: 69964e415698665ec492d852. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"BLIP-2 OPT (6.7B)","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 65.2. Result ID: 69964e415698665ec492d83e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"MiniGPT-4 (7B)","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 64.3. Result ID: 69964e415698665ec492d83a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Gemma-VL (7B)","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### CROSS-ALIGN+ Harmfulness Accuracy: 63.9. Result ID: 69964e415698665ec492d842. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"BLIP-2 OPT (6.7B)","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### Qwen2.5-VL Harmfulness Accuracy: 63.5. Result ID: 69964e415698665ec492d845. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### LLaVA-V1.6 Harmfulness Accuracy: 61.7. Result ID: 69964e415698665ec492d843. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### InternVL2 Harmfulness Accuracy: 59.8. Result ID: 69964e415698665ec492d847. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### DeepSeek-VL-S Harmfulness Accuracy: 58.3. Result ID: 69964e415698665ec492d84b. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### mPLUG-Owl2-Small Harmfulness Accuracy: 58. Result ID: 69964e415698665ec492d84f. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### Qwen2.5-VL Harmfulness Accuracy: 57.8. Result ID: 69964e415698665ec492d835. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### MiniGPT-4 Harmfulness Accuracy: 57. Result ID: 69964e415698665ec492d84d. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### LLaVA-V1.6 Harmfulness Accuracy: 56.9. Result ID: 69964e415698665ec492d833. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### Gemma-VL Harmfulness Accuracy: 56.5. Result ID: 69964e415698665ec492d849. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### InternVL2 Harmfulness Accuracy: 55.9. Result ID: 69964e415698665ec492d837. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### mPLUG-Owl2-Small Harmfulness Accuracy: 55.6. Result ID: 69964e415698665ec492d83f. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### DeepSeek-VL-S Harmfulness Accuracy: 54.9. Result ID: 69964e415698665ec492d83b. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### BLIP-2 OPT Harmfulness Accuracy: 54.9. Result ID: 69964e415698665ec492d851. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"6.7B","Protocol":"Standard In-Context Learning (ICL)"}. This record does not assert independent reproduction. ### MiniGPT-4 Harmfulness Accuracy: 53.8. Result ID: 69964e415698665ec492d83d. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### Gemma-VL Harmfulness Accuracy: 52.6. Result ID: 69964e415698665ec492d839. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"7B","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. ### BLIP-2 OPT Harmfulness Accuracy: 52.3. Result ID: 69964e415698665ec492d841. [Reporting paper]() Recorded evaluation configuration: {"Parameters":"6.7B","Protocol":"Vanilla Zero-Shot Inference"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.