---
language:
- en
tags:
- chemistry
license: cc-by-sa-4.0
datePublished: 2025-05-16
version: 1.0.0
task_categories:
- table-question-answering
- feature-extraction
---
This dataset is part of the Benchmarking Multimodal LLMs on Recognition and Understanding over Chemical Tables project, currently under review at NeurIPS 2025. It includes the full version of the dataset used in our experiments, now publicly available on Hugging Face Datasets to ensure compliance with NeurIPS guidelines regarding accessibility and reproducibility.

The dataset consists of chemical tables extracted from scientific publications, paired with metadata such as DOIs, page-level contexts, and structured annotations. Each sample is associated with its source paper and is designed to support tasks such as table recognition, table understanding, and scientific information extraction in the chemical domain.