UniEvo-VL trains image models to learn from their own mistakes, lifting GenEval 74.7% to 80.8%
mark_k · x · 2026-10-02
A new paper, UniEvo-VL, proposes multimodal self-improvement: the model generates an image, critiques what went wrong, and converts the feedback into corrective instructions. A teacher generator sees those instructions while the student sees only the original prompt; training transfers the correction benefit into the student's weights so improvements persist.
Results (Qwen-Image-2512 with Qwen-VL feedback):
- GenEval rises from 74.7% to 80.8%
- Verifying proposed corrections before training pushes it to 81.8%
- A stronger external critic reaches 88.2%
Outcomes vary across tasks, with mixed results on text rendering. The author highlights the core idea: turning self-critique into lasting improvement — learning to avoid mistakes is far more powerful than spotting them.
More from Multimodal
- Core Memory profiles Ari Kuschnir, the creator behind viral AI parody videos — kyliebytes · 2026-11-19
- AI-generated flat-illustration portraits show off stylized consistency — OVolosin82152 · 2026-11-19
- GPT-6.2 Rigs and Animates a 3D Game Character End-to-End, Dev Puts It Straight Into Production — OpenAIDevs · 2026-11-19
- AssemblyAI Launches STT-3.7 Models With 86ms Agent Latency and Rewritten Inference Engine — AssemblyAI · 2026-11-19
- Pika Enact lets you direct actors and camera inside a running AI video in real time — thetripathi58 · 2026-11-19
- Google DeepMind's Genie 4 adds native sound, chainable worlds, and a community remix gallery — The Decoder · 2026-11-19