| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| HatefulMemes | Qwen2-VL-7B | Accuracy82.1 | 65 | 1mo ago | |
| PrideMM | MemeGuard | Accuracy79.2 | 31 | 3mo ago | |
| MAMI | Qwen2-VL-7B | Accuracy0.799 | 30 | 4mo ago | |
| HarMeme | LLaVA-1.5-7B | Accuracy88.2 | 30 | 4mo ago | |
| Harm-P | Qwen2-VL-7B | Accuracy91.6 | 28 | 4mo ago | |
| MultiOFF | Qwen2-VL-7B | Accuracy71.1 | 27 | 4mo ago | |
| Bn-HIB 1.0 (test) | MCFM | Accuracy77.46 | 17 | 1mo ago | |
| GOAT-C (test) | Accuracy81.5 | 12 | 1mo ago | ||
| MMSD 2.0 (test) | Accuracy86.9 | 12 | 1mo ago | ||
| MemeReaCon CSAC-C (test) | Accuracy71.3 | 12 | 1mo ago | ||
| MemeReaCon CMI-C (test) | Accuracy83.2 | 12 | 1mo ago | ||
| ArMeme (test) | Qwen3-VL-8B-Instruct | Accuracy76.6 | 11 | 1mo ago | |
| HarMeme Dataset (test) | KID-VLM | Accuracy85.03 | 11 | 4mo ago | |
| Tamil Memes (test) | transformer-transformer architecture | Support667 | 5 | 4mo ago |
