Personality Recognition on AVI-6 (test) benchmark leaderboard · SOTA2 Research

Personality Recognition on AVI-6 (test)

0.1555MSE (H)LLM-based AU-text fusion framework

Evaluation Results

Method
LLM-based AU-text fusion framework2026.06	0.1555	0.138	0.1597	0.1077	0.1402	0.3242	0.2833	0.3287	0.2565	0.2982	0.4654	0.7258	0.5154	0.6934	0.6
Longformer (only text)2026.06	0.1815	0.2697	0.2053	0.2097	0.2166	0.3508	0.4009	0.3721	0.3646	0.3721	0.3279	0.2736	0.0617	0.0022	0.1664
Longformer (AU semantic + text)2026.06	0.2015	0.305	0.2058	0.2098	0.2305	0.3632	0.4183	0.3708	0.3613	0.3784	0.12	0.0196	0.0199	0.0533	0.0532
LSTM (Selected AUs)2026.06	0.2057	0.299	0.2188	0.243	0.2416	0.3664	0.4286	0.3858	0.3985	0.3948	-0.0448	0.2236	-0.0343	0.1633	0.077
ViT (SW)2026.06	0.2489	0.3551	0.2882	0.2785	0.2927	0.4084	0.4751	0.4145	0.4203	0.4296	0.1597	-0.041	-0.0131	0.1505	0.064
ResNet + BERT (feature fusion)2026.06	0.256	0.2971	0.528	0.3146	0.3489	0.4036	0.4444	0.5769	0.4507	0.4689	0.1507	0.2573	0.1307	0.2537	0.1981
ResNet (DS)2026.06	0.3743	0.5765	0.4058	0.5048	0.4654	0.4741	0.5987	0.5239	0.5823	0.5448	0.2014	-0.1573	0.1789	-0.0235	0.0499
ResNet (SW)2026.06	0.4583	1.3539	0.9343	0.4982	0.8112	0.5436	0.947	0.8116	0.5777	0.7198	0.0387	-0.04	0.3675	-0.032	0.0836
ResNet + BERT (decision fusion)2026.06	0.6673	0.5503	0.4162	0.4447	0.5196	0.6749	0.6139	0.5166	0.5272	0.5831	0.0215	0.1065	0.0957	0.2397	0.1159
Qwen-3-VL-8B-Instruct2026.06	1.7777	0.4847	1.1193	1.1653	1.1368	1.2245	0.5349	0.9259	0.9574	0.9107	0.3814	0.6683	0.5589	0.5779	0.5467