UMM-Reflection: Interleaved RL Teaches Unified Multimodal Models to Self-Correct Images

Ziqi Huang's team open-sourced UMM-Reflection, which uses interleaved reinforcement learning to teach unified multimodal models to generate, reflect on, and redraw images in a joint loop. The method improves GenEval scores from 0.71 to 0.84.

2026-09-29 ~ 2026-09-29 · 3 related posts

• UMM-Reflection: interleaved RL teaches unified models self-correcting image generation — Yijia Fan · 2026-09-29
• UMM-Reflection: interleaved RL teaches unified models to self-correct, GenEval 0.71→0.84 — ziqi_huang_ · 2026-09-29

1 near-duplicate retellings: ziqi_huang_