Hateful Meme Classification on HarM (test)
91.03AUCPro-CapPromptHate
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Pro-CapPromptHateAugmentation=image entities and demographic information2023.08 | 91.03 | 85.03 | |
| PromptHatetraining_objective=masked language modeling, method=prompting2023.02 | 90.96 | 84.47 | |
| PromptHateAugmentation=image entities and demographic information2023.08 | 90.96 | 84.47 | |
| Pro-CapBERTAugmentation=image entities and demographic information2023.08 | 89.75 | 82.71 | |
| FT-ROBERTatraining_objective=fine-tuning2023.02 | 89.26 | 82.32 | |
| ALBEFAugmentation=image entities and demographic information2023.08 | 86.91 | 81.78 | |
| BLIPAugmentation=image entities and demographic information2023.08 | 86.88 | 82.66 | |
| DisMultiHate2023.02 | 86.39 | 81.24 | |
| DisMultiHateAugmentation=image entities and demographic information2023.08 | 86.39 | 81.24 | |
| MOMENTA2023.02 | 86.32 | 80.48 | |
| MOMENTAAugmentation=image entities and demographic information2023.08 | 86.32 | 80.48 | |
| ISSUES2023.10 | 85.51 | 77.7 | |
| MMBT-Region2023.02 | 85.48 | 79.83 | |
| VILBERT CCAugmentation=image entities and demographic information2023.08 | 84.79 | 81.39 | |
| HMC 1st prize2023.10 | 84.49 | 73.2 | |
| Hate-CLIPper+2023.10 | 84.36 | 77.15 | |
| VILBERT CC2023.02 | 84.11 | 78.7 | |
| CLIP Text + textual inversionmodality=multimodal, textual inversion=true2023.10 | 83.31 | 76.65 | |
| Concat BERT2023.02 | 83.21 | 77.82 | |
| Late Fusion2023.02 | 83.17 | 77.57 | |
| HMC 2nd prize2023.10 | 83.1 | 69.5 | |
| VisualBERT COCOAugmentation=image entities and demographic information2023.08 | 82.96 | 78.81 | |
| CLIP Summodality=multimodal, fusion=sum2023.10 | 82.92 | 76.8 | |
| CLIP BERTText Encoder=BERT, Visual Encoder=CLIP2023.02 | 82.63 | 76.66 | |
| Text BERT2023.02 | 81.39 | 75.68 | |
| CLIP Image-Onlymodality=image-only2023.10 | 81.35 | 74.65 | |
| Visual BERT COCO2023.02 | 80.46 | 75.31 | |
| Image-Region2023.02 | 76.46 | 73.05 | |
| VisualBERTPre-training=COCO2023.10 | 74.59 | 69.95 | |
| ViLBERTPre-training=CC2023.10 | 72.78 | 70.03 | |
| CLIP Text-Onlymodality=text-only2023.10 | 63.43 | 63.5 |