WebGen-Instruct: Training Data for WebGen-Bench

This repository contains WebGen-Instruct, the training data used in the paper WebGen-Bench: Evaluating LLMs on Generating Interactive and Functional Websites from Scratch.

WebGen-Bench is a novel benchmark designed to measure an LLM-based agent's ability to create multi-file website codebases from scratch. The benchmark dataset itself consists of 101 instructions and 647 test cases. This particular dataset (WebGen-Instruct) provides 6,667 website-generation instructions, including 600 trajectories collected from DeepSeek-V3 and filtered by appearance score (larger or equal to 3).

The code for evaluation, as well as the training code and the full WebGen-Bench data, are released at WebGen-Bench (Github).

Sample Usage

You can easily load the training dataset using the load_dataset function from the 🤗 Datasets library:

from datasets import load_dataset

# Load the WebGen-Instruct training dataset
train_dataset = load_dataset("luzimu/WebGen-Bench_train_data", split="train")

# Print dataset information
print(train_dataset)

# Access an example
print(train_dataset[0])

Training Results

The performance of the WebGen-LM models which are trained with this data is shown below:

Citation

If you find our project useful, please cite:

@misc{lu2025webgenbenchevaluatingllmsgenerating,
      title={WebGen-Bench: Evaluating LLMs on Generating Interactive and Functional Websites from Scratch}, 
      author={Zimu Lu and Yunqiao Yang and Houxing Ren and Haotian Hou and Han Xiao and Ke Wang and Weikang Shi and Aojun Zhou and Mingjie Zhan and Hongsheng Li},
      year={2025},
      eprint={2505.03733},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2505.03733}, 
}

@misc{lu2025webgenagentenhancinginteractivewebsite,
      title={WebGen-Agent: Enhancing Interactive Website Generation with Multi-Level Feedback and Step-Level Reinforcement Learning}, 
      author={Zimu Lu and Houxing Ren and Yunqiao Yang and Ke Wang and Zhuofan Zong and Junting Pan and Mingjie Zhan and Hongsheng Li},
      year={2025},
      eprint={2509.22644},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2509.22644}, 
}