ResearchBenchmarksMultimodal Reasoning on Overall Reasoning SuiteFollow63.43Average AccuracyCFPOG62.254862.559962.86563.1701Jun 22, 2026Evaluation ResultsMethodMethodLinksAverage AccuracyCFPOGBackbone=Qwen2.5-VL-7B...Backbone=Qwen2.5-VL-7B, Hyperparameters=gamma=0.02 + No Ent2026.0663.43PAPOGBackbone=Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B2026.0662.3