Why We Feel: Breaking Boundaries in Emotional Reasoning with Multimodal Large Language Models

Benchmark Model Rank Results
emotion-interpretation-on-eibenchClaude-3-haiku#1Recall: 63.24
emotion-interpretation-on-eibenchLLaVA-1.5 (13B)#2Recall: 54.37
emotion-interpretation-on-eibenchLLaVA-NEXT (13B)#3Recall: 54.33
emotion-interpretation-on-eibenchClaude-3-sonnet#4Recall: 54.1
emotion-interpretation-on-eibenchLLaVA-NEXT (7B)#5Recall: 53.82
emotion-interpretation-on-eibenchMiniGPT-v2#6Recall: 52.89
emotion-interpretation-on-eibenchChatGPT-4o#7Recall: 49.99
emotion-interpretation-on-eibenchVideo-LLaVA#8Recall: 49.26
emotion-interpretation-on-eibenchLLaVA-NEXT (34B)#9Recall: 49.03
emotion-interpretation-on-eibenchChatGPT-4V#10Recall: 46.86
emotion-interpretation-on-eibenchOtter#11Recall: 42.81
emotion-interpretation-on-eibenchQwen-vl-plus#12Recall: 31
emotion-interpretation-on-eibenchQwen-VL-Chat#13Recall: 26.45
emotion-interpretation-on-eibench-complexChatGPT-4o#1Recall: 39.27
emotion-interpretation-on-eibench-complexLLaVA-NEXT (13B)#2Recall: 39.16
emotion-interpretation-on-eibench-complexLLaVA-NEXT (7B)#3Recall: 38.71
emotion-interpretation-on-eibench-complexLLaVA-1.5 (13B)#4Recall: 38.1
emotion-interpretation-on-eibench-complexLLaVA-NEXT (34B)#5Recall: 35.37
emotion-interpretation-on-eibench-complexMiniGPT-v2#6Recall: 35.1
emotion-interpretation-on-eibench-complexVideo-LLaVA#7Recall: 30.9
emotion-interpretation-on-eibench-complexChatGPT-4V#8Recall: 28
emotion-interpretation-on-eibench-complexOtter#9Recall: 27.9
emotion-interpretation-on-eibench-complexClaude-3-haiku#10Recall: 24
emotion-interpretation-on-eibench-complexQwen-VL-Chat#11Recall: 22
emotion-interpretation-on-eibench-complexClaude-3-sonnet#12Recall: 21.37
emotion-interpretation-on-eibench-complexQwen-vl-plus#13Recall: 20.37