alexandru-dima
commited on
Commit
•
4627fea
1
Parent(s):
6deffc7
Create new file
Browse files- README.txt +33 -0
README.txt
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
This is a version of squad derived from SQuAD v2 on which we did the following changes:
|
2 |
+
|
3 |
+
def remove_long_answers(example): #MAP
|
4 |
+
new_answers = {'text':[], 'answer_start':[]}
|
5 |
+
answers_set = set()
|
6 |
+
for text, start in zip(example['answers']['text'], example['answers']['answer_start']):
|
7 |
+
if len(text.split()) > 10:
|
8 |
+
continue
|
9 |
+
else:
|
10 |
+
if (text, start) not in answers_set:
|
11 |
+
new_answers['text'].append(text)
|
12 |
+
new_answers['answer_start'].append(start)
|
13 |
+
answers_set.add((text, start))
|
14 |
+
example['answers'] = new_answers
|
15 |
+
return example
|
16 |
+
|
17 |
+
def filter_whitespace(example): #MAP
|
18 |
+
example['context'] = ' '.join(example['context'].split())
|
19 |
+
example['question'] = ' '.join(example['question'].split())
|
20 |
+
example['answers']['text'] = [' '.join(answer.split()) for answer in example['answers']['text']]
|
21 |
+
return example
|
22 |
+
|
23 |
+
def checks(example): #FILTER
|
24 |
+
if example['answers']['text'] == []:
|
25 |
+
return False
|
26 |
+
|
27 |
+
if len(example['question'].split()) > 35:
|
28 |
+
return False
|
29 |
+
|
30 |
+
if len(example['question'].split()) < 3:
|
31 |
+
return False
|
32 |
+
|
33 |
+
return True
|