unigrpo-unified-visual-generation-rl

【设计与创意】Replace classifier-free guidance and KL-based regularization in flow matching with velocity-based MSE regularization and GRPO for joint text-image optimization. Achieves 0.8381 TextAlign and 0.90 GenEval without CFG overhead. Works best for multimodal generation where text and image must stay coherent. Trigger: When optimizing vision-language models …

独立 GitHub 快照 · 自动分类 · 以原作者文档为准