Share your thoughts, 1 month free Claude Pro on usSee more

Multimodal Reasoning on MathVision (Metrics: FEI, AEI)

46.07FEI

VRPRM-MiMo

Updated 12d ago

Evaluation Results

Method	Links
VRPRM-MiMo 2025.08		46.07	72.87
VRPRM-MiMo w/o CoT 2025.08		43.4	69.99
VRPRM 2025.08		42.98	67.34
VRPRM-Qwen3 2025.08		42.13	70.48
VRPRM-Qwen3 w/o CoT 2025.08		39.89	70.15
VRPRM-MiMo w/o RL 2025.08		39.36	60.51
MiMo-VL-7B 2025.08		38.27	61.6
VRPRM-Qwen3 w/o RL 2025.08		37.78	57.69
Qwen3-VL-4B-Thinking 2025.08		37.08	61.64
Qwen2.5-VL-7B 2025.08		36.94	39.54
VRPRM-MiMo w/o RL & w/o CoT 2025.08		35.81	44.24
VRPRM-Qwen3 w/o RL & w/o CoT 2025.08		34.97	53.49
Qwen2.5-VL-72B 2025.08		34.27	41.88
VRPRM w/o RL 2025.08		33.99	61.82
VRPRM w/o RL & w/o CoT 2025.08		33.15	54.57
InternVL2.5-8B 2025.08		30.67	42.61
Gemini-2.0-Flash 2025.08		30.48	40.68
GPT-4o-mini 2025.08		27.39	35.1
VRPRM w/o CoT 2025.08		26.83	51.95
VisualPRM-8B 2025.08		24.58	62.91