Multi-modal Stance Detection on MTSE
72.68DT Macro F1GPT-4 Vision
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-4 VisionModality=Multi-modal, Zero-shot=true2026.04 | 72.68 | 71.28 | |
| GPT-4 + CoTModality=Textual, Zero-shot=true2026.04 | 71.05 | 69.1 | |
| MV-DebateModality=Textual, Zero-shot=true2026.04 | 70.92 | 69.33 | |
| GPT-4Modality=Textual, Zero-shot=true2026.04 | 70.78 | 68.83 | |
| GPT-4 VisionModality=Multi-modal, Evaluation Protocol=in-target2026.04 | 70.46 | 72.82 | |
| MM-StanceDetModality=Multi-modal, Evaluation Protocol=in-target2026.04 | 70.12 | 73.66 | |
| MV-DebateModality=Textual, Evaluation Protocol=in-target2026.04 | 69.45 | 66.91 | |
| BridgeTowerModality=Multi-modal, Zero-shot=true2026.04 | 69.15 | 69.88 | |
| GPT-4 + CoTModality=Textual, Evaluation Protocol=in-target2026.04 | 69.12 | 67.05 | |
| GPT-4Modality=Textual, Evaluation Protocol=in-target2026.04 | 68.74 | 66.39 | |
| BridgeTowerModality=Multi-modal, Evaluation Protocol=in-target2026.04 | 68.53 | 71.02 | |
| TASTEModality=Multi-modal, Evaluation Protocol=in-target2026.04 | 68.14 | 68.52 | |
| MM-StanceDetModality=Multi-modal, Zero-shot=true2026.04 | 67.21 | 71.03 | |
| LKI-BARTModality=Textual, Evaluation Protocol=in-target2026.04 | 65.23 | 70.11 | |
| KEBERTModality=Textual, Evaluation Protocol=in-target2026.04 | 64.5 | 69.81 | |
| TASTEModality=Multi-modal, Zero-shot=true2026.04 | 62.34 | 63.01 | |
| RoBERTaModality=Textual, Evaluation Protocol=in-target2026.04 | 58.39 | 60.79 | |
| TMPTModality=Multi-modal, Evaluation Protocol=in-target2026.04 | 55.41 | 61.61 | |
| LLaMA2Modality=Textual, Zero-shot=true2026.04 | 53.57 | 53.92 | |
| LLaMA2Modality=Textual, Evaluation Protocol=in-target2026.04 | 53.23 | 52.67 | |
| CLIPModality=Multi-modal, Evaluation Protocol=in-target2026.04 | 53.22 | 65.83 | |
| BERTModality=Textual, Evaluation Protocol=in-target2026.04 | 48.25 | 52.04 | |
| Qwen-VLModality=Multi-modal, Zero-shot=true2026.04 | 47.62 | 46.14 | |
| Qwen-VLModality=Multi-modal, Evaluation Protocol=in-target2026.04 | 43.31 | 45.13 | |
| BERT+ViTModality=Multi-modal, Evaluation Protocol=in-target2026.04 | 41.86 | 45.82 | |
| ViTModality=Visual, Evaluation Protocol=in-target2026.04 | 40.48 | 40.42 | |
| SwinTModality=Visual, Evaluation Protocol=in-target2026.04 | 39.89 | 40.43 | |
| ResNetModality=Visual, Evaluation Protocol=in-target2026.04 | 37.89 | 38.59 | |
| ViLTModality=Multi-modal, Evaluation Protocol=in-target2026.04 | 35.32 | 48.24 | |
| BERTModality=Textual, Zero-shot=true2026.04 | 32.52 | 29.97 | |
| TMPTModality=Multi-modal, Zero-shot=true2026.04 | 31.69 | 32.65 | |
| ViTModality=Visual, Zero-shot=true2026.04 | 28.63 | 29.7 | |
| SwinTModality=Visual, Zero-shot=true2026.04 | 28.54 | 30.85 | |
| CLIPModality=Multi-modal, Zero-shot=true2026.04 | 28.21 | 28.99 | |
| ViLTModality=Multi-modal, Zero-shot=true2026.04 | 28.08 | 29.74 | |
| BERT+ViTModality=Multi-modal, Zero-shot=true2026.04 | 26.7 | 31.57 | |
| RoBERTaModality=Textual, Zero-shot=true2026.04 | 26.6 | 32.21 | |
| LKI-BARTModality=Textual, Zero-shot=true2026.04 | 26.53 | 32.11 | |
| KEBERTModality=Textual, Zero-shot=true2026.04 | 26.17 | 31.81 | |
| ResNetModality=Visual, Zero-shot=true2026.04 | 25.52 | 29.7 |