Binary Hate Detection on H-VLI (Overall)
88.06AccuracyGPT-5
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| GPT-5Training=False, Framework=ARCADE2026.03 | 88.06 | 85.96 | 84.2 | |
| GPT-4oTraining=False, Framework=ARCADE2026.03 | 84.98 | 86.43 | 81.78 | |
| GPT-4.1-miniTraining=False, Framework=ARCADE2026.03 | 84.73 | 83.37 | 80.98 | |
| GPT-5-miniTraining=False, Framework=ARCADE2026.03 | 84.47 | 94.75 | 82.63 | |
| Gemini-2.5-ProTraining=False, Framework=ARCADE2026.03 | 83.02 | 86.21 | 79.84 | |
| Gemini-2.5-FlashTraining=False, Framework=ARCADE2026.03 | 82.42 | 87.31 | 79.48 | |
| Gemini-2.5-ProTraining=False2026.03 | 81.92 | 93.07 | 80.15 | |
| Qwen3-VL-PlusTraining=False, Framework=ARCADE2026.03 | 80.84 | 82.89 | 77.14 | |
| Gemini-2.5-FlashTraining=False2026.03 | 79.97 | 92.42 | 78.35 | |
| GPT-4.1-miniTraining=False2026.03 | 77.59 | 84.85 | 74.81 | |
| Qwen-VL-MaxTraining=False, Framework=ARCADE2026.03 | 77.35 | 77.46 | 72.76 | |
| GPT-4oTraining=False2026.03 | 76.4 | 94.37 | 75.83 | |
| GPT-5Training=False2026.03 | 74.32 | 68.25 | 64.56 | |
| Qwen3-VL-PlusTraining=False2026.03 | 73.85 | 80.92 | 70.69 | |
| Qwen-VL-PlusTraining=False, Framework=ARCADE2026.03 | 73.52 | 61.01 | 64.19 | |
| BERT+ViTTraining=True2026.03 | 72.67 | 71 | 67.08 | |
| MMBTTraining=True2026.03 | 72.5 | 68.18 | 66.04 | |
| BERTTraining=True2026.03 | 71.48 | 68.83 | 65.43 | |
| PromptHateTraining=True2026.03 | 69.27 | 71.86 | 64.72 | |
| Qwen-VL-MaxTraining=False2026.03 | 69.11 | 90.83 | 69.68 | |
| GPT-5-miniTraining=False2026.03 | 67.57 | 77.92 | 65.34 | |
| Qwen-VL-PlusTraining=False2026.03 | 67.32 | 80.35 | 65.77 | |
| MomentaTraining=True2026.03 | 66.55 | 65.37 | 60.52 | |
| ViT-b-16Training=True2026.03 | 53.06 | 40.69 | 40.47 |