List your company
ResearchFormerly WizWand

Fine-grained Hate Categorization on H-VLI (Normal)

88.64AccuracyGPT-5
6.16827.57948.9970.401Mar 22, 2026
Updated 7mo ago

Evaluation results

MethodLinks
GPT-5Training=False, Framework=ARCADE2026.0388.64
GPT-4.1-miniTraining=False, Framework=ARCADE2026.0382.99
MMBTTraining=True2026.0380.28
BERT+ViTTraining=True2026.0377.51
Qwen3-VL-PlusTraining=False, Framework=ARCADE2026.0377.43
MomentaTraining=True2026.0376.47
Gemini-2.5-FlashTraining=False, Framework=ARCADE2026.0374.31
PromptHateTraining=True2026.0373.36
GPT-4oTraining=False, Framework=ARCADE2026.0373.26
Gemini-2.5-ProTraining=False, Framework=ARCADE2026.0373.26
BERTTraining=True2026.0372.66
GPT-5-miniTraining=False, Framework=ARCADE2026.0370.83
ViT-b-16Training=True2026.0368.86
Qwen-VL-MaxTraining=False, Framework=ARCADE2026.0365.97
Qwen-VL-PlusTraining=False, Framework=ARCADE2026.0361.46
GPT-5Training=False2026.0355.56
Gemini-2.5-FlashTraining=False2026.0355.02
Gemini-2.5-ProTraining=False2026.0352.94
GPT-4oTraining=False2026.0336.33
Qwen3-VL-PlusTraining=False2026.0334.26
GPT-5-miniTraining=False2026.0329.76
GPT-4.1-miniTraining=False2026.0328.72
Qwen-VL-PlusTraining=False2026.0310.03
Qwen-VL-MaxTraining=False2026.039.34