medical_pretrain / README.md
DavidLanz's picture
Update README.md
d6cd52a
---
license: apache-2.0
language:
- zh
- en
tags:
- text-generation
pretty_name: medical
task_categories:
- text-generation
size_categories:
- 1M<n<10M
---
### Description
- **medical_pretrain_tw.json**: This dataset contains a total of 360,000 entries sourced from medical encyclopedia data from FreedomIntelligence/huatuo_encyclopedia_qa. These entries are a combination of questions and answers, forming text fields with coherent sentences. The dataset is intended for pre-training purposes to inject medical knowledge.
- **medical_book_zh.json**: This dataset includes 8,475 entries sourced from text data in medical textbooks. The data source is [here](https://github.com/jind11/MedQA), and the original dataset was obtained from Google Drive. It has been processed to split long paragraphs into small sections, each containing a maximum of 2048 characters.
### Data Format
- **medical_pretrain_tw.json.json**: JSON format with text fields.
- **medical_book_zh.json**: JSON format with text fields.
### License
Please refer to the respective data sources for licensing information.
### Dataset Citation
If you use this dataset in your research or work, please consider citing the original data sources as specified above.