Text and Image To Text Vision models
Compare text+image-to-text (vision) models side-by-side and evaluate them on factuality, instruction following, helpfulness, style, and overall preference. Comparing two responses and deciding which handles the prompt more effectively.