Pref-GRPO: Pairwise Preference Reward-based GRPO for Stable Text-to-Image Reinforcement Learning

Documenti analoghi

• Fine-Grained GRPO for Precise Preference Alignment in Flow Models
  di: Zhou, Yujie, et al.
  Pubblicazione: (2025)
• UniGenBench++: A Unified Semantic Evaluation Benchmark for Text-to-Image Generation
  di: Wang, Yibin, et al.
  Pubblicazione: (2025)
• Unified Multimodal Chain-of-Thought Reward Model through Reinforcement Fine-Tuning
  di: Wang, Yibin, et al.
  Pubblicazione: (2025)
• Pave-GRPO: Beyond Instantaneous Guidance through Principled Average Velocity Decomposition
  di: Ling, Pengyang, et al.
  Pubblicazione: (2026)
• Unified Personalized Reward Model for Vision Generation
  di: Wang, Yibin, et al.
  Pubblicazione: (2026)