Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2604.10918 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917402042695680 |
|---|---|
| author | Yang, Yunfan Lan, Cuiling Sang, Jitao Lu, Yan |
| author_facet | Yang, Yunfan Lan, Cuiling Sang, Jitao Lu, Yan |
| contents | Tables contain rich structured information, yet when stored as images their contents remain "locked" within pixels. Converting table images into LaTeX code enables faithful digitization and reuse, but current multimodal large language models (MLLMs) often fail to preserve structural, style, or content fidelity. Conventional post-training with reinforcement learning (RL) typically relies on a single aggregated reward, leading to reward ambiguity that conflates multiple behavioral aspects and hinders effective optimization. We propose Component-Specific Policy Optimization (CSPO), an RL framework that disentangles optimization across LaTeX tables components-structure, style, and content. In particular, CSPO assigns component-specific rewards and backpropagates each signal only through the tokens relevant to its component, alleviating reward ambiguity and enabling targeted component-wise optimization. To comprehensively assess performance, we introduce a set of hierarchical evaluation metrics. Extensive experiments demonstrate the effectiveness of CSPO, underscoring the importance of component-specific optimization for reliable structured generation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_10918 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CSPO: Alleviating Reward Ambiguity for Structured Table-to-LaTeX Generation Yang, Yunfan Lan, Cuiling Sang, Jitao Lu, Yan Artificial Intelligence Tables contain rich structured information, yet when stored as images their contents remain "locked" within pixels. Converting table images into LaTeX code enables faithful digitization and reuse, but current multimodal large language models (MLLMs) often fail to preserve structural, style, or content fidelity. Conventional post-training with reinforcement learning (RL) typically relies on a single aggregated reward, leading to reward ambiguity that conflates multiple behavioral aspects and hinders effective optimization. We propose Component-Specific Policy Optimization (CSPO), an RL framework that disentangles optimization across LaTeX tables components-structure, style, and content. In particular, CSPO assigns component-specific rewards and backpropagates each signal only through the tokens relevant to its component, alleviating reward ambiguity and enabling targeted component-wise optimization. To comprehensively assess performance, we introduce a set of hierarchical evaluation metrics. Extensive experiments demonstrate the effectiveness of CSPO, underscoring the importance of component-specific optimization for reliable structured generation. |
| title | CSPO: Alleviating Reward Ambiguity for Structured Table-to-LaTeX Generation |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2604.10918 |