AI RESEARCH
OP-GRPO: Efficient Off-Policy GRPO for Flow-Matching Models
arXiv CS.CV
•
ArXi:2604.04142v1 Announce Type: new