What to Edit Next: Visually Aligned Image-Editing Follow-Up Suggestions in Conversational Systems
The episode discusses a paper titled "What to Edit Next: Visually Aligned Image-Editing Follow-Up Suggestions in Conversational Systems." The hosts detail a three-stage framework used by researchers to improve image editing suggestions by combining supervised fine-tuning, reinforcement learning based on user clicks, and a visual verifier. The system achieved significant improvements in click-through rate and reduced visual inconsistency.