Search papers, labs, and topics across Lattice.
This paper introduces RubricRM, a novel pairwise generative reward modeling framework designed to enhance the evaluation of visual generative models by employing input-specific rubrics that adapt to various instructions. By utilizing a two-stage training pipeline that combines supervised fine-tuning with dimension-level rewards, RubricRM achieves superior performance in both text-to-image generation and image editing tasks compared to existing reward models. The results demonstrate that RubricRM not only surpasses specialized reward models but also competes effectively with proprietary MLLM judges, all while leveraging smaller model architectures.
RubricRM adapts evaluation criteria dynamically, leading to significant performance gains in visual generative tasks compared to static reward models.
Reward models play an essential role in aligning visual generative models, yet most existing visual reward models use a single scalar score or rely on fixed criteria that cannot adapt to different instructions. This limits both interpretability and task sensitivity, especially for text-to-image generation and instruction-based image editing, where different inputs require different evaluation dimensions. We propose RubricRM, a pairwise generative reward modeling framework that first produces an input-specific rubric with evaluation dimensions, weights, and scoring criteria, and then applies the rubric to score candidate images. We train dedicated RubricRM models for text-to-image generation and image editing using a two-stage training pipeline: supervised fine-tuning teaches the model the rubric-based scoring paradigm, while GRPO further improves scoring through fine-grained dimension-level rewards. Experiments on multiple generation and editing benchmarks show that RubricRM outperforms existing specialized reward models and remains competitive with strong proprietary MLLM judges despite using smaller backbones. Our models, data, and code are available at https://github.com/zijiankan/RubricRM.
