PEGASUS

anchor

Findings

• IC-900BLIP-2, MiniGPT-4, and LLaVA-1.5 show degraded zero-shot VQA accuracy on underspecified questions, with absolute improvements of 1.14–7.94% when questions are augmented with visually-grounded details [compared-to]
• IC-902BLIP-2 and MiniGPT-4 confidence-based question selection underperforms the original question for paraphrased candidates but succeeds for semantically enriched REPARe questions [compared-to]