AnyDoc: Enhancing Document Generation via Large-Scale HTML/CSS Data Synthesis and Height-Aware Reinforcement Optimization

TL;DR

AnyDoc introduces a large-scale HTML/CSS dataset and a height-aware reinforcement learning method to improve multi-task document generation, achieving superior results over existing models.

Contribution

The paper presents a scalable data synthesis pipeline, a large multi-category dataset DocHTML, and a height-aware reinforcement learning approach for enhanced document generation.

Findings

AnyDoc outperforms baseline models on all tasks.

The dataset covers 111 categories and 32 styles.

Height-aware reinforcement learning reduces content overflow.

Abstract

Document generation has gained growing attention in the field of AI-driven content creation. In this work, we push its boundaries by introducing AnyDoc, a framework capable of handling multiple generation tasks across a wide spectrum of document categories, all represented in a unified HTML/CSS format. To overcome the limited coverage and scale of existing human-crafted document datasets, AnyDoc first establishes a scalable data synthesis pipeline to automatically generate documents in HTML/CSS form. This pipeline yields DocHTML, a large-scale dataset containing 265,206 document samples, while spanning 111 categories and 32 distinct styles. Additionally, all documents are equipped with comprehensive metadata, including design intentions, HTML/CSS source code, visual assets, and rendered screenshots. Building on the curated dataset, AnyDoc fine-tunes multi-modal large language models (MLLMs) to achieve three practical document generation tasks: intention-to-document, document derendering, and element-to-document. To address the content overflow issue observed during fine-tuning, AnyDoc further incorporates a height-aware reinforcement learning (HARL) post-training procedure. By defining a reward function based on the difference between predicted and target document heights, overflow is penalized and gradually mitigated during HARL, thereby enhancing overall performance. Qualitative and quantitative experiments demonstrate that AnyDoc outperforms both general-purpose MLLMs and task-specific baselines across all three tasks.