Muennighoff
commited on
Commit
β’
37c1004
1
Parent(s):
eb2ad9a
Fixes
Browse files- README.md +10 -0
- raw_data/convert_tsv.py +33 -0
- {data β raw_data}/xwinograd.tsv +0 -0
- test/{en.json β en.jsonl} +0 -0
- test/{fr.json β fr.jsonl} +0 -0
- test/{jp.json β jp.jsonl} +0 -0
- test/{pt.json β pt.jsonl} +0 -0
- test/{ru.json β ru.jsonl} +0 -0
- test/{zh.json β zh.jsonl} +0 -0
- xwinograd.py +3 -6
README.md
CHANGED
@@ -1,3 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
## XWinograd
|
2 |
|
3 |
Multilingual winograd schema challenge.
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- en
|
4 |
+
- fr
|
5 |
+
- jp
|
6 |
+
- pt
|
7 |
+
- ru
|
8 |
+
- zh
|
9 |
+
---
|
10 |
+
|
11 |
## XWinograd
|
12 |
|
13 |
Multilingual winograd schema challenge.
|
raw_data/convert_tsv.py
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import json
|
2 |
+
import random
|
3 |
+
|
4 |
+
def winogrande_format(row):
|
5 |
+
array = row["pronoun"]
|
6 |
+
position_idx = json.loads(array)[1][0]
|
7 |
+
# Turn unicode into proper chinese characters
|
8 |
+
sent = str(u"{}".format(row["sent"]))
|
9 |
+
start_idx = 0
|
10 |
+
for i, tok in enumerate(json.loads(row["toks"])):
|
11 |
+
tok = str(u"{}".format(tok))
|
12 |
+
cur_start_idx = sent.find(tok)
|
13 |
+
if i == position_idx:
|
14 |
+
break
|
15 |
+
sent = sent[cur_start_idx + len(tok):]
|
16 |
+
start_idx += cur_start_idx + len(tok)
|
17 |
+
# +1 to give room for an optional space
|
18 |
+
row["sentence"] = row["sent"][:start_idx] + row["sent"][start_idx:start_idx+len(tok)+1].replace(tok, "_") + row["sent"][start_idx+len(tok)+1:]
|
19 |
+
|
20 |
+
sol = json.loads(row["solution"])
|
21 |
+
|
22 |
+
cor_answer_idx = random.choice([1, 2])
|
23 |
+
incor_answer_idx = 2 if cor_answer_idx == 1 else 1
|
24 |
+
|
25 |
+
cor_answer = str(u"{}".format(sol[0][0])) if sol[0][-1] == True else str(u"{}".format(sol[1][0]))
|
26 |
+
incor_answer = str(u"{}".format(sol[0][0])) if sol[0][-1] == False else str(u"{}".format(sol[1][0]))
|
27 |
+
|
28 |
+
row[f"option{cor_answer_idx}"] = cor_answer
|
29 |
+
row[f"option{incor_answer_idx}"] = incor_answer
|
30 |
+
row["answer"] = cor_answer_idx
|
31 |
+
return row
|
32 |
+
|
33 |
+
# ds = ds.apply(winogrande_format, axis=1)
|
{data β raw_data}/xwinograd.tsv
RENAMED
File without changes
|
test/{en.json β en.jsonl}
RENAMED
File without changes
|
test/{fr.json β fr.jsonl}
RENAMED
File without changes
|
test/{jp.json β jp.jsonl}
RENAMED
File without changes
|
test/{pt.json β pt.jsonl}
RENAMED
File without changes
|
test/{ru.json β ru.jsonl}
RENAMED
File without changes
|
test/{zh.json β zh.jsonl}
RENAMED
File without changes
|
xwinograd.py
CHANGED
@@ -1,10 +1,7 @@
|
|
1 |
"""XWinograd"""
|
2 |
|
3 |
-
|
4 |
import json
|
5 |
|
6 |
-
import pandas as pd
|
7 |
-
|
8 |
import datasets
|
9 |
|
10 |
|
@@ -61,7 +58,7 @@ class XWinograd(datasets.GeneratorBasedBuilder):
|
|
61 |
|
62 |
def _split_generators(self, dl_manager):
|
63 |
|
64 |
-
downloaded_files =
|
65 |
return [
|
66 |
datasets.SplitGenerator(
|
67 |
name=datasets.Split.TEST,
|
@@ -71,14 +68,14 @@ class XWinograd(datasets.GeneratorBasedBuilder):
|
|
71 |
|
72 |
def _generate_examples(self, filepath):
|
73 |
"""This function returns the examples in the raw (text) form."""
|
74 |
-
logger.info("
|
75 |
|
76 |
with open(filepath, encoding="utf-8") as f:
|
77 |
for id_, row in enumerate(f):
|
78 |
data = json.loads(row)
|
79 |
|
80 |
yield id_, {
|
81 |
-
"sentence": data["
|
82 |
"option1": data["option1"],
|
83 |
"option2": data["option2"],
|
84 |
"answer": data["answer"],
|
|
|
1 |
"""XWinograd"""
|
2 |
|
|
|
3 |
import json
|
4 |
|
|
|
|
|
5 |
import datasets
|
6 |
|
7 |
|
|
|
58 |
|
59 |
def _split_generators(self, dl_manager):
|
60 |
|
61 |
+
downloaded_files = dl_manager.download(_URL.format(lang=self.config.name))
|
62 |
return [
|
63 |
datasets.SplitGenerator(
|
64 |
name=datasets.Split.TEST,
|
|
|
68 |
|
69 |
def _generate_examples(self, filepath):
|
70 |
"""This function returns the examples in the raw (text) form."""
|
71 |
+
logger.info("Generating examples from = %s", filepath)
|
72 |
|
73 |
with open(filepath, encoding="utf-8") as f:
|
74 |
for id_, row in enumerate(f):
|
75 |
data = json.loads(row)
|
76 |
|
77 |
yield id_, {
|
78 |
+
"sentence": data["sentence"],
|
79 |
"option1": data["option1"],
|
80 |
"option2": data["option2"],
|
81 |
"answer": data["answer"],
|