Vision-Language Tasks

This research examines the use of GPT-4V as an automated evaluator for vision-language tasks. The paper validates GPT-4V’s capabilities for evaluation purposes, addressing tasks from basic image-to-text and text-to-image synthesis…