README.md

---
license: apache-2.0
task_categories:
- image-to-text
language:
- en
- zh
tags:
- agent
- code
size_categories:
- 100K<n<1M
---

<p align="center">
    <img src="./docs/assets/logo.svg" alt="Logo" width="120" />
    <p align="center">
        <a href="https://github.com/PKU-DAIR">
            <img alt="Static Badge" src="https://img.shields.io/badge/%C2%A9-PKU--DAIR-%230e529d?labelColor=%23003985">
        </a>
    </p>
</p>

## **WebRenderBench: Enhancing Web Interface Generation through Layout-Style Consistency and Reinforcement Learning**

[Paper](https://arxiv.org/pdf/2510.04097) | [中文](./docs/Chinese.md)

## **🔍 Overview**

**WebRenderBench** is a large-scale benchmark designed to advance **WebUI-to-Code** research for multimodal large language models (MLLMs) through evaluation on real-world webpages. It provides:

* **45,100** real webpages collected from public portal websites
* **High diversity and complexity**, covering a wide range of industries and design styles
* **Novel evaluation metrics** that quantify **layout and style consistency** based on rendered pages
* The **ALISA reinforcement learning framework**, which uses the new metrics as reward signals to optimize generation quality

---

## **🚀 Key Features**

### **Beyond the Limitations of Traditional Benchmarks**

WebRenderBench addresses the core issues of existing WebUI-to-Code benchmarks in data quality and evaluation methodology:

| Aspect                     | Traditional Benchmarks                                                                          | Advantages of WebRenderBench                                                                 |
| :------------------------- | :---------------------------------------------------------------------------------------------- | :------------------------------------------------------------------------------------------- |
| **Data Quality**           | Small-scale, simple-structured, or LLM-synthesized data with limited diversity                  | Large-scale, real-world, and structurally complex webpages that present higher challenges    |
| **Evaluation Reliability** | Relies on visual APIs (high cost) or code-structure comparison (fails to handle code asymmetry) | Objectively and efficiently evaluates layout and style consistency based on rendered results |
| **Training Effectiveness** | Difficult to optimize on crawled data with asymmetric code structures                           | Proposed metrics can be directly used as RL reward signals to enhance model optimization     |

---

### **Dataset Characteristics**

<p align="center">  
<img src="./docs/assets/framework.svg" alt="WebRenderBench and ALISA Framework" width="80%" />  
</p>  
<p align="center"><i>Figure 1: Dataset construction pipeline and the ALISA framework</i></p>  

Our dataset is constructed through a systematic process to ensure both **high quality** and **diversity**:

1. **Data Collection**: URLs are obtained from open enterprise portal datasets. A high-concurrency crawler captures 210K webpages along with static resources.
2. **Data Processing**: MHTML pages are converted into HTML files, and cross-domain resources are processed to ensure local renderability and full-page screenshots.
3. **Data Cleaning**: Pages with abnormal sizes, rendering errors, or missing styles are filtered out. Multimodal QA models further remove low-quality samples with large blank areas or overlapping elements, yielding 110K valid pages.
4. **Data Categorization**: Pages are categorized by industry and complexity (measured via *Group Count*) to ensure balanced distribution across difficulty levels and domains.

Finally, we construct a dataset of **45.1K** samples, evenly split into training and test sets.

---

## **🌟 Evaluation Framework**

We propose a novel evaluation protocol based on **rendered webpages**, quantifying model performance along two key dimensions: **layout** and **style consistency**.

---

### **RDA (Relative Layout Difference of Associated Elements)**

**Purpose:** Measures relative layout differences between matched elements.

* **Element Association:** Matches corresponding elements between generated and target pages using text similarity (LCS) and geometric distance.
* **Positional Deviation:** The page is divided into a 3×3 grid. Associated elements are compared quadrant-wise—if located in different quadrants, the score is 0; otherwise, a deviation-based score is computed.
* **Uniqueness Weighting:** Each element is weighted by its uniqueness (inverse group size), giving higher importance to distinctive components.

---

### **GDA (Group-wise Difference in Element Counts)**

**Purpose:** Measures group-level alignment of axis-aligned elements.

* **Grouping:** Elements aligned on the same horizontal or vertical axis are treated as one group.
* **Count Comparison:** Compares whether corresponding groups in the generated and target pages contain the same number of elements.
* **Uniqueness Weighting:** Weighted by element uniqueness to emphasize key structural alignment.

---

### **SDA (Style Difference of Associated Elements)**

**Purpose:** Evaluates fine-grained style differences between associated elements.

* **Multi-Dimensional Style Extraction:** Measures differences in foreground color, background color, font size, and border radius.
* **Weighted Averaging:** Computes a weighted mean of style similarity scores across all associated elements to obtain an overall style score.

---

## **⚙️ Installation Guide**

### **Core Dependencies**

<!--  
# Recommended: Use vLLM for faster inference  
pip install vllm transformers>=4.40.0 torch>=2.0

# Other dependencies  
pip install selenium pandas scikit-learn pillow

Alternatively:
pip install -r requirements.txt  
-->

Coming Soon

---

## **📊 Benchmark Workflow**

### **Directory Structure**

```
|- docs/                     # Documentation
|- scripts                   # Evaluation scripts  
|- web_render_test.jsonl      # Test set metadata  
|- web_render_train.jsonl     # Training set metadata  
|- test_webpages.zip          # Test set webpages  
|- train_webpages.zip         # Training set webpages  
|- test_screenshots.zip       # Test set screenshots  
|- train_screenshots.zip      # Training set screenshots  
```

---

### **Obtain Datasets**

- Webpages

| File Name | Download Link (ModelScope) |
|--------|---------------------|
| train_webpages.7z.001 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.001) |
| train_webpages.7z.002 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.002) |
| train_webpages.7z.003 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.003) |
| train_webpages.7z.004 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.004) |
| train_webpages.7z.005 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.005) |
| train_webpages.7z.006 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.006) |
| train_webpages.7z.007 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.007) |
| train_webpages.7z.008 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.008) |
| train_webpages.7z.009 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.009) |
| train_webpages.7z.010 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.010) |
| train_webpages.7z.011 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.011) |
| train_webpages.7z.012 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.012) |
| train_webpages.7z.013 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.013) |
| train_webpages.7z.014 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.014) |
| train_webpages.7z.015 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.015) |
| train_webpages.7z.016 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.016) |
| train_webpages.7z.017 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.017) |
| train_webpages.7z.018 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.018) |
| train_webpages.7z.019 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_webpages.7z.019) |
| test_webpages.7z.001 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.001) |
| test_webpages.7z.002 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.002) |
| test_webpages.7z.003 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.003) |
| test_webpages.7z.004 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.004) |
| test_webpages.7z.005 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.005) |
| test_webpages.7z.006 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.006) |
| test_webpages.7z.007 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.007) |
| test_webpages.7z.008 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.008) |
| test_webpages.7z.009 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.009) |
| test_webpages.7z.010 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.010) |
| test_webpages.7z.011 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.011) |
| test_webpages.7z.012 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.012) |
| test_webpages.7z.013 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.013) |
| test_webpages.7z.014 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.014) |
| test_webpages.7z.015 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.015) |
| test_webpages.7z.016 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.016) |
| test_webpages.7z.017 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.017) |
| test_webpages.7z.018 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_webpages.7z.018) |

- Screenshots

| File Name | Download Link (ModelScope) |
|--------|---------------------|
| train_screenshots.7z.001 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_screenshots.7z.001) |
| train_screenshots.7z.002 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/train_screenshots.7z.002) |
| test_screenshots.7z.001 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_screenshots.7z.001) |
| test_screenshots.7z.002 | [Download](https://www.modelscope.cn/datasets/lpc1290/WebRenderBench/resolve/master/test_screenshots.7z.002) |

### **Implementation Steps**

1. **Data Preparation**

   * Download the WebRenderBench dataset and extract webpage and screenshot archives.
   * Each pair consists of a real webpage (HTML + resources) and its rendered screenshot.

2. **Model Inference**

   * Run inference using engines such as **vLLM** or **LLM Deploy**, and save results to the designated directory.

3. **Evaluation**

   * Run `scripts/1_get_evaluation.py`.
   * The script launches a web server to render both generated and target HTML.
   * WebDriver extracts DOM information and computes **RDA**, **GDA**, and **SDA** scores.
   * Results are saved under `save_results/`.
   * Final scores are aggregated via `scripts/2_compute_alisa_scores.py`.

4. **ALISA Training (Optional)**

   * Use `models/train_rl.py` for reinforcement learning fine-tuning. *(Coming Soon)*
   * The computed evaluation scores serve as reward signals to optimize policy models via methods such as **GRPO**.

---

## **📈 Model Performance Insights**

We evaluate **17 multimodal large language models** of varying scales and architectures (both open- and closed-source).

* **Combined Scores of RDA, GDA, and SDA (%)**

![Inference Results](./docs/assets/inference_results.png)

**Key Findings:**

* Overall, larger models achieve higher consistency. **GPT-4.1-mini** and **Qwen-VL-Plus** perform best among closed-source models.
* While most models perform reasonably on simple pages (*Group Count* < 50), **RDA scores drop sharply** as page complexity increases—precise layout alignment remains a major challenge.
* After reinforcement learning via the **ALISA framework**, **Qwen2.5-VL-7B** shows substantial improvements across all complexity levels, even surpassing **GPT-4.1-mini** on simpler cases.

---

## **📅 Future Work**

* [ ] Release pretrained models fine-tuned with the ALISA framework
* [ ] Expand dataset coverage to more industries and dynamic interaction patterns
* [ ] Open-source the complete toolchain for data collection, cleaning, and evaluation

---

## **📜 License**

The **WebRenderBench dataset** is released for **research purposes only**.
All accompanying code will be published under the **Apache License 2.0**.

All webpages in the dataset are collected from publicly accessible enterprise portals.
To protect privacy, all personal and sensitive information has been removed or modified.

---

## **📚 Citation**

If you use our dataset or framework in your research, please cite the following paper:

```bibtex
@article{webrenderbench2025,
  title={WebRenderBench: Enhancing Web Interface Generation through Layout-Style Consistency and Reinforcement Learning},
  author={Anonymous Author(s)},
  year={2025},
  journal={arXiv preprint},
}
```