Multi-modal Stance Detection on MWTWT in-target benchmark leaderboard · SOTA2 Research

Multi-modal Stance Detection on MWTWT in-target

76.6CA Macro F1TMPT

Evaluation Results

Method	Links
TMPT2026.04		76.6	63.19	67.25	62.92	81.19
BERT2026.04		75.62	60.85	63.05	59.24	81.53
LKI-BART2026.04		72.01	67.8	69.54	69.92	80.83
KEBERT2026.04		71.67	67.56	69.29	69.74	80.57
TASTE2026.04		71.55	69.23	70.88	71.91	71.93
MM-StanceDet2026.04		71.49	68.3	71.93	66.5	67.76
CLIP2026.04		70.93	67.17	67.43	70.86	79.06
BridgeTower2026.04		70.18	68.31	67.92	65.44	79.58
RoBERTa2026.04		69.56	65.03	69.74	67.99	79.21
MV-Debate2026.04		64.22	66.03	69.87	72.31	52.95
GPT-4 + CoT2026.04		64.01	65.88	70.1	72.05	53.11
BERT+ViT2026.04		63.2	44.71	56.45	46.85	73.71
GPT-42026.04		63.14	65.12	69.93	71.62	52.69
ViLT2026.04		62.7	56.44	58.06	60.22	73.66
ViT2026.04		46.63	50	40.16	46.32	50.86
SwinT2026.04		46.3	46.99	41.02	47.39	51.32
GPT-4 Vision2026.04		44.59	47.07	57.47	57.9	37.61
Qwen-VL2026.04		43.06	45.49	49.79	46.04	27.73
ResNet2026.04		39.89	42.2	43.52	37.05	50.34
LLaMA22026.04		34.89	41.95	49.09	44.32	30.21