# Hateful Meme Detection on MAMI

Source: [Hateful Meme Detection on MAMI](<https://www.sota2.com/research/sota/hateful-meme-detection-on-mami>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-04T11:59:00.107Z

Evaluation of model performance on the MAMI dataset for Hateful Meme Detection, typically measured using AUC and Accuracy.

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 | Hateful Meme Detection |
| Dataset (including any stated split) | MAMI |
| Metric | F1 (Binary) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 29 of 29 |
| 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

- [F1 (Binary)](<https://www.sota2.com/research/sota/hateful-meme-detection-on-mami.md?metric=F1%20(Binary)>): Higher is better; unit: PERCENT.

- [AUC](<https://www.sota2.com/research/sota/hateful-meme-detection-on-mami.md?metric=AUC>): Higher is better; unit: SCALAR.

- [Accuracy](<https://www.sota2.com/research/sota/hateful-meme-detection-on-mami.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [F1 (Attack)](<https://www.sota2.com/research/sota/hateful-meme-detection-on-mami.md?metric=F1%20(Attack)>): Higher is better; unit: PERCENT.

- [Reasoning Score (CDE)](<https://www.sota2.com/research/sota/hateful-meme-detection-on-mami.md?metric=Reasoning%20Score%20(CDE)>): Higher is better; unit: SCALAR.

## Recorded results

| Method | F1 (Binary) | Result date | Report paper |
| --- | --- | --- | --- |
| ExPO-HM | 82.3 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| RA-HMD | 81 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| ExPO-HM | 80.7 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| U-CoT+ (14B) | 79.9 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| SFT | 78.1 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| SFT | 77.9 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| GRPO | 76.8 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| GRPO | 76.6 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| LOREHM (34B) | 75.3 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| SFT | 72.8 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| DPO | 72.3 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| SFT | 69.2 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| DPO | 66.8 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| Zero-shot | 63.9 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| Zero-shot | 63.4 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| Zero-shot | 61.1 | Not recorded | 68fa7bfd1e209131cea8bb3e |
| Zero-shot | 58.7 | Not recorded | 68fa7bfd1e209131cea8bb3e |

### ExPO-HM

F1 (Binary): 82.3. Result ID: 69a81e726dbb7075b94e1ebe.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems","Backbone":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction.

### RA-HMD

F1 (Binary): 81. Result ID: 69a81e726dbb7075b94e1eb2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b681e209131cea883ce.md>)

[Implementation](<https://github.com/JingbiaoMei/RGCL>)

Recorded evaluation configuration: {"System Category":"Direct Detection Baselines"}. This record does not assert independent reproduction.

### ExPO-HM

F1 (Binary): 80.7. Result ID: 69a81e726dbb7075b94e1eb9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems","Backbone":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction.

### U-CoT+ (14B)

F1 (Binary): 79.9. Result ID: 69a81e726dbb7075b94e1eb4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa793d1e209131cea7cca3.md>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems"}. This record does not assert independent reproduction.

### SFT

F1 (Binary): 78.1. Result ID: 69a81e726dbb7075b94e1eb1.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

Recorded evaluation configuration: {"System Category":"Direct Detection Baselines","Backbone":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction.

### SFT

F1 (Binary): 77.9. Result ID: 69a81e726dbb7075b94e1eaf.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

Recorded evaluation configuration: {"System Category":"Direct Detection Baselines","Backbone":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction.

### GRPO

F1 (Binary): 76.8. Result ID: 69a81e726dbb7075b94e1ebd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems","Backbone":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction.

### GRPO

F1 (Binary): 76.6. Result ID: 69a81e726dbb7075b94e1eb8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems","Backbone":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction.

### LOREHM (34B)

F1 (Binary): 75.3. Result ID: 69a81e726dbb7075b94e1eb3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75961e209131cea68cb5.md>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems"}. This record does not assert independent reproduction.

### SFT

F1 (Binary): 72.8. Result ID: 69a81e726dbb7075b94e1ebb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems","Backbone":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction.

### DPO

F1 (Binary): 72.3. Result ID: 69a81e726dbb7075b94e1ebc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems","Backbone":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction.

### SFT

F1 (Binary): 69.2. Result ID: 69a81e726dbb7075b94e1eb6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems","Backbone":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction.

### DPO

F1 (Binary): 66.8. Result ID: 69a81e726dbb7075b94e1eb7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems","Backbone":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction.

### Zero-shot

F1 (Binary): 63.9. Result ID: 69a81e726dbb7075b94e1eba.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems","Backbone":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction.

### Zero-shot

F1 (Binary): 63.4. Result ID: 69a81e726dbb7075b94e1eb0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

Recorded evaluation configuration: {"System Category":"Direct Detection Baselines","Backbone":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction.

### Zero-shot

F1 (Binary): 61.1. Result ID: 69a81e726dbb7075b94e1eae.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

Recorded evaluation configuration: {"System Category":"Direct Detection Baselines","Backbone":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction.

### Zero-shot

F1 (Binary): 58.7. Result ID: 69a81e726dbb7075b94e1eb5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7bfd1e209131cea8bb3e.md>)

Recorded evaluation configuration: {"System Category":"Explain-then-Detect Systems","Backbone":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
