Reasoning-Based Rewards for Image Editing

2026-05-04

Edit-R1 introduces a chain-of-thought reward model that evaluates image edits through structured reasoning. This model aims to enhance alignment and performance in text-guided image editing tasks.