WebRenderBench: Enhancing Web Interface Generation through Layout-Style Consistency and Reinforcement Learning
This work addresses the need for more reliable and diverse benchmarks in web interface generation, which is important for front-end development and rapid prototyping, though it appears incremental in building on existing multimodal LLM approaches.
The authors tackled the problem of automating UI image-to-code conversion by introducing WebRenderBench, a large-scale benchmark with 45.1k webpages, and ALISA, a reinforcement learning agent that improves generation performance to achieve state-of-the-art results.
Automating the conversion of UI images into web code is a critical task for front-end development and rapid prototyping. Advances in multimodal large language models (MLLMs) have made WebUI-to-Code increasingly feasible, yet existing benchmarks remain limited in data diversity and evaluation reliability. To address these issues, we present WebRenderBench, a large-scale benchmark of 45.1k webpages collected from real-world portal sites, offering greater diversity, complexity, and realism than prior benchmarks. We further propose a novel evaluation metric that measures layout and style consistency from the final rendered pages. Unlike vision-based methods that rely on costly LLM reasoning or structure-based comparisons vulnerable to noise and asymmetry, our approach enables more efficient, objective, and reliable UI quality assessment. Finally, we introduce the Automated Layout and Style Inspection Agent (ALISA), which integrates this metric into reinforcement learning as a reward signal to enhance training on crawled asymmetric webpages. Experiments show that ALISA significantly boosts generation performance, achieving state-of-the-art results across multiple metrics.