---
license: apache-2.0
---

This is the _cleaned_ version of Deepset's [COVID-QA dataset](https://aclanthology.org/2020.nlpcovid19-acl.18/) which is described in section 3.2 of [Towards Efficient Methods in Medical Question Answering using Knowledge Graph Embeddings](https://arxiv.org/abs/2401.07977). While you can use either version of the dataset, we recommend using this version since we have corrected many mistakes in the original question set & also, it yields better scores, even zero-shot, as reported in our paper.