ResearchBenchmarksVisual Question Answering on DocVQA (test)Visual Question Answering on DocVQA (test)Follow95.9AccuracyEMOVA86.12488.66291.293.738Sep 26, 2024Evaluation resultsMethodMethodLinksAccuracyEMOVAModel Size=72BModel Size=72B2024.0995.9EMOVAModel Size=7BModel Size=7B2024.0994.2EMOVAModel Size=3BModel Size=3B2024.0993.5GPT-4o2024.0992.8GPT-4V2024.0988.4Gemini Pro 1.52024.0986.5