Datasets:

Gholamreza
/

pquad

@@ -68,14 +68,13 @@ dataset_info:
 ---
 # Dataset Card for "pquad"
-## PQuAD
-*THIS IS A NON-OFFICIAL VERSION OF THE DATASET UPLOADED TO HUGGINGFACE BY [Gholamreza Dar](https://huggingface.co/Gholamreza)*
-The original repository for the dataset is https://github.com/AUT-NLP/PQuAD
 Original README.md:
 PQuAD is a crowd- sourced reading comprehension dataset on Persian Language. It includes 80,000
 questions along with their answers, with 25% of the questions being unanswerable. As a reading
 comprehension dataset, it requires a system to read a passage and then answer the given questions
@@ -83,6 +82,7 @@ from the passage. PQuAD's questions are based on Persian Wikipedia articles and
 variety of subjects. Articles used for question generation are quality checked and include few
 number of non-Persian words.
 The dataset is divided into three categories including train, validation, and test sets and the
 statistics of these sets are as follows:
@@ -108,6 +108,7 @@ PQuAD is stored in the JSON format and consists of passages where each passage i
 set of questions. Answer(s) of the questions is specified with answer's span (start and end
 point of answer in paragraph). Also, the unanswerable questions are marked as unanswerable.
 The estimated human performance on the test set is 88.3% for F1 and 80.3% for EM. We have
 evaluated PQuAD using two pre-trained transformer-based language models, namely ParsBERT
 (Farahani et al., 2021) and XLM-RoBERTa (Conneau et al., 2020), as well as BiDAF (Levy et
@@ -124,6 +125,7 @@ al., 2017) which is an attention-based model proposed for MRC.
 +-------------+------+------+-----------+-----------+-------------+
 ```
 PQuAD is developed by Mabna Intelligent Computing at Amirkabir Science and Technology Park with
 collaboration of the NLP lab of the Amirkabir University of Technology and is supported by the
 Vice Presidency for Scientific and Technology. By releasing this dataset, we aim to ease research

 ---
 # Dataset Card for "pquad"
+## PQuAD Description
+**THIS IS A NON-OFFICIAL VERSION OF THE DATASET UPLOADED TO HUGGINGFACE BY [Gholamreza Dar](https://huggingface.co/Gholamreza)**
+*The original repository for the dataset is https://github.com/AUT-NLP/PQuAD*
 Original README.md:
 PQuAD is a crowd- sourced reading comprehension dataset on Persian Language. It includes 80,000
 questions along with their answers, with 25% of the questions being unanswerable. As a reading
 comprehension dataset, it requires a system to read a passage and then answer the given questions
 variety of subjects. Articles used for question generation are quality checked and include few
 number of non-Persian words.
+## Dataset Splits
 The dataset is divided into three categories including train, validation, and test sets and the
 statistics of these sets are as follows:
 set of questions. Answer(s) of the questions is specified with answer's span (start and end
 point of answer in paragraph). Also, the unanswerable questions are marked as unanswerable.
+##Results
 The estimated human performance on the test set is 88.3% for F1 and 80.3% for EM. We have
 evaluated PQuAD using two pre-trained transformer-based language models, namely ParsBERT
 (Farahani et al., 2021) and XLM-RoBERTa (Conneau et al., 2020), as well as BiDAF (Levy et
 +-------------+------+------+-----------+-----------+-------------+
 ```
+##LICENSE
 PQuAD is developed by Mabna Intelligent Computing at Amirkabir Science and Technology Park with
 collaboration of the NLP lab of the Amirkabir University of Technology and is supported by the
 Vice Presidency for Scientific and Technology. By releasing this dataset, we aim to ease research