Abstract
Precise instruction following in image generation, such as satisfying object counts and spatial relations, remains an open challenge at least in part because it is learned using unreliable reward models such as object detectors and vision-language models. We introduce Verifiable Visual Rewards (VVR), the first framework for programmatically verifiable image rewards, and show that training on it generalizes to natural prompts. Each VVR task is a scene of geometric objects and relations among them, from which we derive both the prompt and a deterministic verifier, so tasks can be generated in any number and at any chosen complexity. We release VVRBench, with 10,000 tasks over 32 constraint types, and VVRBench-Challenge, with 720 more complex tasks; the strongest model we evaluate---GPT-Image-2.5---solves 21.4% of VVRBench-Challenge. Using VVR scores as rewards for reinforcement learning (RLVVR) raises the accuracy of Stable Diffusion 3.5 Medium on VVRBench from 2.8% to 28.3% and demonstrates consistent easy-to-hard generalization. These gains extend to out-of-domain benchmarks, and mixing VVR into existing objectives further improves overall performance and human preference, motivating the adoption of VVR into standard image generation post-training recipes.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Li, S. S., Han, X., Tsvetkov, Y., & Zettlemoyer, L. (2026). Verifiable Visual Rewards Transfer from Synthetic Scenes to Natural Prompts. https://omanscience.com/en/articles/verifiable-visual-rewards-transfer-from-synthetic-scenes-to-natural-prompts
MLA 9
Li, Shuyue Stella, et al. "Verifiable Visual Rewards Transfer from Synthetic Scenes to Natural Prompts." https://omanscience.com/en/articles/verifiable-visual-rewards-transfer-from-synthetic-scenes-to-natural-prompts.
Chicago (author–date)
Li, Shuyue Stella, Xiaochuang Han, Yulia Tsvetkov, and Luke Zettlemoyer. 2026. "Verifiable Visual Rewards Transfer from Synthetic Scenes to Natural Prompts." https://omanscience.com/en/articles/verifiable-visual-rewards-transfer-from-synthetic-scenes-to-natural-prompts.
Harvard
Li, S. S., Han, X., Tsvetkov, Y. and Zettlemoyer, L. (2026) 'Verifiable Visual Rewards Transfer from Synthetic Scenes to Natural Prompts', Available at: https://omanscience.com/en/articles/verifiable-visual-rewards-transfer-from-synthetic-scenes-to-natural-prompts.
Vancouver
Li SS, Han X, Tsvetkov Y, Zettlemoyer L. Verifiable Visual Rewards Transfer from Synthetic Scenes to Natural Prompts. https://omanscience.com/en/articles/verifiable-visual-rewards-transfer-from-synthetic-scenes-to-natural-prompts
IEEE
S. S. Li, X. Han, Y. Tsvetkov, and L. Zettlemoyer, "Verifiable Visual Rewards Transfer from Synthetic Scenes to Natural Prompts," https://omanscience.com/en/articles/verifiable-visual-rewards-transfer-from-synthetic-scenes-to-natural-prompts.