Update README.md

README.md

@@ -29,3 +29,16 @@ configs:
   - split: val
     path: data/val-*
 ---
+
+# arXiv Abstract
+
+This dataset is based on the arXiv scientific papers and is used for the text expansion task. ([Download raw data here](https://drive.google.com/file/d/1b3rmCSIoh6VhD4HKWjI4HOW-cSwcwbeC/view?usp=sharing)).
+
+I processed the raw data for the article expansion task with `extract_arXiv_abstract.py`. The processed dataset only contains the article ID and abstract fields. The JSON objects are in the following format:
+
+```
+{ 
+  'article_id': str,
+  'abstract_text': List[str],
+  'token_count': int
+}