Convert dataset to Parquet

Convert dataset to Parquet.

This dataset uses `tasks`, which are deprecated and will raise an error after the next release of `datasets`. See: https://github.com/huggingface/datasets/pull/6999

README.md

@@ -18,9 +18,9 @@ task_categories:
 task_ids:
 - closed-domain-qa
 - extractive-qa
-paperswithcode_id: null
 pretty_name: COVID-QA
 dataset_info:
+  config_name: covid_qa_deepset
   features:
   - name: document_id
     dtype: int32
@@ -38,13 +38,18 @@ dataset_info:
       dtype: string
     - name: answer_start
       dtype: int32
-  config_name: covid_qa_deepset
   splits:
   - name: train
-    num_bytes: 65151262
+    num_bytes: 65151242
     num_examples: 2019
-  download_size: 4418117
-  dataset_size: 65151262
+  download_size: 2274275
+  dataset_size: 65151242
+configs:
+- config_name: covid_qa_deepset
+  data_files:
+  - split: train
+    path: covid_qa_deepset/train-*
+  default: true
 ---
 
 

covid_qa_deepset/train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1662a83af1bd1fc14bf86af1c60386ad8c836ec26896b189765ffe0c92836df
+size 2274275