Complete UniGRPO implementation/config for MMaDA

[hongjx175]

Great work!
I found only MMU settings in the mmada_v example configs. Could you please let me know if there is a full UniGRPO implementation or corresponding complete configuration available in this repo? Thanks for your work.

[yinjjiew]

Thanks for your interest in this work. This repo only contains basic reinforcement learnining methods, not the inplementation of UniGRPO in MMaDA's paper. If your want to aggreate multiple reward together, I think you only need to change ./reward/rl_reward_v.py.