GT-SVJ: Generative-Transformer-Based Self-Supervised Video Judge For Efficient Video Reward Modeling
This addresses the problem of inefficient reward modeling for video generation in AI research, offering a more data-efficient solution.
The paper tackles the challenge of aligning video generative models with human preferences by proposing a novel approach that repurposes video generative models as temporally-aware reward models, achieving state-of-the-art performance on GenAI-Bench and MonteBench with only 30K human annotations—6× to 65× fewer than existing methods.
Aligning video generative models with human preferences remains challenging: current approaches rely on Vision-Language Models (VLMs) for reward modeling, but these models struggle to capture subtle temporal dynamics. We propose a fundamentally different approach: repurposing video generative models, which are inherently designed to model temporal structure, as reward models. We present the Generative-Transformer-based Self-Supervised Video Judge (\modelname), a novel evaluation model that transforms state-of-the-art video generation models into powerful temporally-aware reward models. Our key insight is that generative models can be reformulated as energy-based models (EBMs) that assign low energy to high-quality videos and high energy to degraded ones, enabling them to discriminate video quality with remarkable precision when trained via contrastive objectives. To prevent the model from exploiting superficial differences between real and generated videos, we design challenging synthetic negative videos through controlled latent-space perturbations: temporal slicing, feature swapping, and frame shuffling, which simulate realistic but subtle visual degradations. This forces the model to learn meaningful spatiotemporal features rather than trivial artifacts. \modelname achieves state-of-the-art performance on GenAI-Bench and MonteBench using only 30K human-annotations: $6\times$ to $65\times$ fewer than existing VLM-based approaches.