Paper, models, ablations + single-reward specialists, and the live ZeroGPU demo.
-
MIRO: MultI-Reward cOnditioned pretraining improves T2I quality and efficiency
Paper • 2510.25897 • Published • 17 -
nicolas-dufour/miro
Text-to-Image • 0.4B • Updated • 606 • 14 -
nicolas-dufour/miro-ablations
Text-to-Image • Updated -
MIRO
🎨6Multi-reward conditioned text-to-image diffusion (ICML 2026)