ResearchBenchmarksLong-form Image Captioning on CapMAS IIW-400Follow46.78ALOHaILVAD43.639244.454645.2746.0854May 20, 2026Evaluation ResultsMethodMethodLinksALOHaFactualityCoverageCLAIRAverage ScoreILVADBackbone=LLaVA-1.5-7B,...Backbone=LLaVA-1.5-7B, Evaluator=GPT-4o2026.0546.7865.4534.851.949.73VHRBackbone=LLaVA-1.5-7B,...Backbone=LLaVA-1.5-7B, Evaluator=GPT-4o2026.0544.0562.4925.854.846.79GreedyBackbone=LLaVA-1.5-7B,...Backbone=LLaVA-1.5-7B, Evaluator=GPT-4o2026.0543.7662.7833.350.747.64