pretty_name: TCM-Ladder
license: cc-by-4.0
language:
- zh
task_categories:
- question-answering
- image-classification
size_categories:
- 1K<n<10K
TCM-Ladder: A Benchmark for Multimodal Question Answering on Traditional Chinese Medicine
TCM-Ladder is the first multimodal QA dataset specifically designed for evaluating large TCM language models. The dataset spans multiple core disciplines of TCM, including fundamental theory, diagnostics, herbal formulas, internal medicine, surgery, pharmacognosy, and pediatrics.
Dataset Details
Dataset Description
TCM-Ladder is the first multimodal QA dataset specifically designed for evaluating large TCM language models. The dataset spans multiple core disciplines of TCM, including fundamental theory, diagnostics, herbal formulas, internal medicine, surgery, pharmacognosy, and pediatrics. In addition to textual content, TCM-Ladder incorporates various modalities such as images and videos. The datasets were constructed using a combination of automated and manual filtering processes and comprise 52,000+ questions in total. These questions include single-choice, multiple-choice, fill-in-the-blank, diagnostic dialogue, and visual comprehension tasks.