CREward: A Type-Specific Creativity Reward Model

2025-11-25

AI Summary

Conventional creativity evaluation treats creativity as a single scalar value, neglecting its inherently multidimensional nature. Method: We propose CREward, the first type-specific creativity reward model for images, quantifying creativity along three interpretable dimensions—geometry, material, and texture. CREward is trained via supervised multi-task learning on a human-perception benchmark augmented with pseudo-labels from large vision-language models (VLMs). It is then integrated as a reward signal to guide LoRA-based fine-tuning for creative image generation. Contribution/Results: CREward significantly outperforms baselines in cross-type creativity assessment, enables fine-grained attribution analysis, and improves both diversity and human preference scores of generated images. To our knowledge, this is the first work to achieve type-decoupled creativity modeling and end-to-end optimization of creative image generation, establishing a new paradigm for controllable creative synthesis.

Technology Category

• Cognitive Modeling & Cognitive Systems: Computational Creativity
• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Diffusion Models for Vision

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Creativity is a complex phenomenon. When it comes to representing and assessing creativity, treating it as a single undifferentiated quantity would appear naive and underwhelming. In this work, we learn the emph{first type-specific creativity reward model}, coined CREward, which spans three creativity ``axes," geometry, material, and texture, to allow us to view creativity through the lens of the image formation pipeline. To build our reward model, we first conduct a human benchmark evaluation to capture human perception of creativity for each type across various creative images. We then analyze the correlation between human judgments and predictions by large vision-language models (LVLMs), confirming that LVLMs exhibit strong alignment with human perception. Building on this observation, we collect LVLM-generated labels to train our CREward model that is applicable to both evaluation and generation of creative images. We explore three applications of CREward: creativity assessment, explainable creativity, and creative sample acquisition for both human design inspiration and guiding creative generation through low-rank adaptation.

Problem

• Model assesses creativity across geometry, material, texture types
• Aligns vision-language model predictions with human creativity perception
• Enables creativity evaluation and generation for image applications

Innovation

• Type-specific creativity reward model across three axes
• Leveraging LVLM-human alignment for scalable training data
• Enabling creativity assessment and generation through low-rank adaptation