← back to paper
arxiv: 2606.05950 · 2 revisions
Edit-R2: Context-Aware Reinforcement Learning for Multi-Turn Image Editing