Update files from the datasets library (from 1.7.0)

Release notes: https://github.com/huggingface/datasets/releases/tag/1.7.0

README.md

@@ -18,6 +18,7 @@ task_categories:
 task_ids:
 - open-domain-qa
 - extractive-qa
+paperswithcode_id: covidqa
 ---
 
 
@@ -26,12 +27,12 @@ task_ids:
 ## Table of Contents
 - [Dataset Description](#dataset-description)
   - [Dataset Summary](#dataset-summary)
-  - [Supported Tasks](#supported-tasks-and-leaderboards)
+  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
   - [Languages](#languages)
 - [Dataset Structure](#dataset-structure)
   - [Data Instances](#data-instances)
-  - [Data Fields](#data-instances)
-  - [Data Splits](#data-instances)
+  - [Data Fields](#data-fields)
+  - [Data Splits](#data-splits)
 - [Dataset Creation](#dataset-creation)
   - [Curation Rationale](#curation-rationale)
   - [Source Data](#source-data)

covid_qa_castorini.py

@@ -96,9 +96,9 @@ class CovidQaCastorini(datasets.GeneratorBasedBuilder):
         logger.info("generating examples from = %s", filepath)
         with open(filepath, encoding="utf-8") as f:
             covid_qa = json.load(f)
-            for article in covid_qa["categories"]:
+            for article_idx, article in enumerate(covid_qa["categories"]):
                 category_name = article["name"]
-                for idx, paragraph in enumerate(article["sub_categories"]):
+                for paragraph_idx, paragraph in enumerate(article["sub_categories"]):
                     question_query = paragraph["nq_name"]
                     keyword_query = paragraph["kq_name"]
 
@@ -106,7 +106,7 @@ class CovidQaCastorini(datasets.GeneratorBasedBuilder):
                     titles = [answer["title"] for answer in paragraph["answers"]]
                     exact_answers = [answer["exact_answer"] for answer in paragraph["answers"]]
 
-                    yield idx, {
+                    yield f"{article_idx}_{paragraph_idx}", {
                         "category_name": category_name,
                         "question_query": question_query,
                         "keyword_query": keyword_query,