Summary of Tcmbench: a Comprehensive Benchmark For Evaluating Large Language Models in Traditional Chinese Medicine, by Wenjing Yue and Xiaoling Wang and Wei Zhu and Ming Guan and Huanran Zheng and Pengfei Wang and Changzhi Sun and Xin Ma
TCMBench: A Comprehensive Benchmark for Evaluating Large Language Models in Traditional Chinese Medicine
by Wenjing Yue, Xiaoling Wang, Wei Zhu, Ming Guan, Huanran Zheng, Pengfei Wang, Changzhi Sun, Xin Ma
First submitted to arxiv on: 3 Jun 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed paper introduces TCM-Bench, a comprehensive benchmark for evaluating Large Language Models (LLMs) in Traditional Chinese Medicine (TCM). The benchmark consists of the TCM-ED dataset, which includes 5,473 questions sourced from the TCM Licensing Exam. To assess LLM performance beyond question accuracy, TCMScore is proposed as a metric that considers consistency with TCM semantics and knowledge. Experimental analyses reveal unsatisfactory LLM performance on this benchmark, highlighting their room for improvement in TCM. Fine-tuning processes can enhance performance, but may also affect basic LLM capabilities. Traditional metrics like Rouge and BertScore are limited by text length and surface semantic ambiguity, while domain-specific metrics like TCMScore provide a more profound evaluation. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Large language models have done well in many tasks, including medical ones. However, they haven’t been tested much for understanding traditional Chinese medicine (TCM). To fill this gap, researchers created a benchmark called TCM-Bench. This includes a dataset of questions from the TCM Licensing Exam and a way to score how well LLMs answer those questions. The results show that LLMs are not very good at answering these questions yet, but they can do better with more training. This is important because it could help make LLMs better tools for medical research. |
Keywords
» Artificial intelligence » Fine tuning » Rouge » Semantics
