# Dataset Card

## Dataset Summary

We split [the original xquad dataset] (https://github.com/deepmind/xquad) into subsets.
We keep the original data format.

## Supported Tasks
extractive question answering

## Language
Thai

## Dataset Split
There are 876/161/153 question-answer pairs from 34/7/7 articles for train/validation/test separately.