ResearchBenchmarksMultimodal Reasoning on MMMU-Pro(V) (Accuracy)Follow37.23AccuracyCFPOG35.087635.643836.236.7562Jun 22, 2026Evaluation ResultsMethodMethodLinksAccuracyCFPOGBackbone=Qwen2.5-VL-7B...Backbone=Qwen2.5-VL-7B, Hyperparameters=gamma=0.02 + No Ent2026.0637.23PAPOGBackbone=Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B2026.0636.76GRPOBackbone=Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B2026.0635.17