update

README.md

@@ -51,7 +51,6 @@ The data fields are the same among all splits.
 ## Citation Information
 
 ```
-
 @article{2016arXiv160605250R,
        author = {{Rajpurkar}, Pranav and {Zhang}, Jian and {Lopyrev},
                  Konstantin and {Liang}, Percy},
@@ -63,5 +62,4 @@ The data fields are the same among all splits.
 archivePrefix = {arXiv},
        eprint = {1606.05250},
 }
-
 ```

process.py

@@ -0,0 +1,26 @@
+import os
+import json
+from datasets import load_dataset
+
+os.makedirs('datasets', exist_ok=True)
+
+data = load_dataset("lmqg/qg_squad")
+_id = 0
+for _split in data:
+    output = []
+    for d in data[_split]:
+        a = d['answer']
+        p = d['paragraph']
+        output.append({
+            "id": str(_id),
+            "title": "None",
+            "context": d['paragraph'],
+            "question": d['question'],
+            "answers": {
+                "text": [a],
+                "answer_start": [p.index(a)]
+            }
+        })
+        _id += 1
+    with open(f'datasets/{_split}.jsonl', 'w') as f:
+        f.write('\n'.join([json.dumps(i) for i in output]))

qa_squad.py

@@ -8,27 +8,21 @@ _VERSION = "0.0.0"
 _NAME = "qa_squad"
 _DESCRIPTION = """SQuAD with the train/validation/test split used in SQuAD QG"""
 _CITATION = """
-@inproceedings{du-cardie-2018-harvesting,
-    title = "Harvesting Paragraph-level Question-Answer Pairs from {W}ikipedia",
-    author = "Du, Xinya  and
-      Cardie, Claire",
-    booktitle = "Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
-    month = jul,
-    year = "2018",
-    address = "Melbourne, Australia",
-    publisher = "Association for Computational Linguistics",
-    url = "https://aclanthology.org/P18-1177",
-    doi = "10.18653/v1/P18-1177",
-    pages = "1907--1917",
-    abstract = "We study the task of generating from Wikipedia articles question-answer pairs that cover content beyond a single sentence. We propose a neural network approach that incorporates coreference knowledge via a novel gating mechanism. As compared to models that only take into account sentence-level information (Heilman and Smith, 2010; Du et al., 2017; Zhou et al., 2017), we find that the linguistic knowledge introduced by the coreference representation aids question generation significantly, producing models that outperform the current state-of-the-art. We apply our system (composed of an answer span extraction system and the passage-level QG system) to the 10,000 top ranking Wikipedia articles and create a corpus of over one million question-answer pairs. We provide qualitative analysis for the this large-scale generated corpus from Wikipedia.",
+@article{2016arXiv160605250R,
+       author = {{Rajpurkar}, Pranav and {Zhang}, Jian and {Lopyrev},
+                 Konstantin and {Liang}, Percy},
+        title = "{SQuAD: 100,000+ Questions for Machine Comprehension of Text}",
+      journal = {arXiv e-prints},
+         year = 2016,
+          eid = {arXiv:1606.05250},
+        pages = {arXiv:1606.05250},
+archivePrefix = {arXiv},
+       eprint = {1606.05250},
 }
 """
-_BASE_URL = "https://huggingface.co/datasets/lmqg/qa_squad/resolve/main/dataset"
-_URLS = {
-    str(datasets.Split.TRAIN): f'{_BASE_URL}/train.json',
-    str(datasets.Split.VALIDATION): f'{_BASE_URL}/dev.json',
-    str(datasets.Split.TEST): f'{_BASE_URL}/test.json'
-}
+_BASE_URL = "https://huggingface.co/datasets/lmqg/qa_squad/resolve/main/datasets"
+_URLS = {k: f'{_BASE_URL}/{k}.json' for k in
+         [str(datasets.Split.TEST), str(datasets.Split.TRAIN), str(datasets.Split.VALIDATION)]}
 
 
 class QASquadConfig(datasets.BuilderConfig):
@@ -104,4 +98,4 @@ class QASquad(datasets.GeneratorBasedBuilder):
                                 "text": answers,
                             },
                         }
-                        key += 1
+                        key += 1