AI RESEARCH
Finite Difference Flow Optimization for RL Post-Training of Text-to-Image Models
arXiv CS.AI
•
ArXi:2603.12893v1 Announce Type: cross Reinforcement learning (RL) has become a standard technique for post-