Self-Rewarding Vision-Language Model via Reasoning Decomposition Paper • 2508.19652 • Published Aug 27, 2025 • 84