Sebastian Gehrmann
commited on
Commit
•
22c2550
1
Parent(s):
1a9b728
add train/val/test
Browse files
xsum.py
CHANGED
@@ -17,12 +17,12 @@ _DESCRIPTION = """\
|
|
17 |
This is the XSUM subset of the GEM benchmark.
|
18 |
"""
|
19 |
_URLs = {
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
|
27 |
_XSUM_REMOVE_LINES = set(
|
28 |
[
|
@@ -40,6 +40,7 @@ _XSUM_REMOVE_LINES = set(
|
|
40 |
]
|
41 |
)
|
42 |
|
|
|
43 |
class Xsum(datasets.GeneratorBasedBuilder):
|
44 |
|
45 |
BUILDER_CONFIGS = [
|
@@ -53,7 +54,7 @@ class Xsum(datasets.GeneratorBasedBuilder):
|
|
53 |
def _info(self):
|
54 |
return datasets.DatasetInfo(
|
55 |
description=_DESCRIPTION,
|
56 |
-
|
57 |
{
|
58 |
"gem_id": datasets.Value("string"),
|
59 |
"gem_parent_id": datasets.Value("string"),
|
@@ -61,8 +62,8 @@ class Xsum(datasets.GeneratorBasedBuilder):
|
|
61 |
"document": datasets.Value("string"),
|
62 |
"target": datasets.Value("string"),
|
63 |
"references": [datasets.Value("string")],
|
64 |
-
}
|
65 |
-
|
66 |
supervised_keys=None,
|
67 |
homepage="",
|
68 |
citation=_CITATION,
|
@@ -75,12 +76,43 @@ class Xsum(datasets.GeneratorBasedBuilder):
|
|
75 |
("challenge_train_sample", "train_xsum_RandomSample500.json"),
|
76 |
("challenge_validation_sample", "validation_xsum_RandomSample500.json"),
|
77 |
("challenge_test_backtranslation", "test_xsum_BackTranslation500.json"),
|
78 |
-
(
|
79 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
80 |
("challenge_test_nopunc", "test_xsum_WithoutPunctuation500.json"),
|
81 |
-
("challenge_test_covid", f"en_test_covid19.jsonl"),
|
82 |
]
|
83 |
return [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
84 |
datasets.SplitGenerator(
|
85 |
name=challenge_split,
|
86 |
gen_kwargs={
|
@@ -88,7 +120,7 @@ class Xsum(datasets.GeneratorBasedBuilder):
|
|
88 |
"split": challenge_split,
|
89 |
},
|
90 |
)
|
91 |
-
for challenge_split, filename in challenge_sets
|
92 |
]
|
93 |
|
94 |
def _generate_examples(self, filepath, split, filepaths=None):
|
@@ -121,9 +153,15 @@ class Xsum(datasets.GeneratorBasedBuilder):
|
|
121 |
with open(filepath, "r", encoding="utf-8") as f:
|
122 |
split_ids = json.load(f)
|
123 |
for id_, i in enumerate(split_ids[split]):
|
124 |
-
with open(
|
|
|
|
|
125 |
text = "".join(
|
126 |
-
[
|
|
|
|
|
|
|
|
|
127 |
)
|
128 |
segs = text.split("[SN]")
|
129 |
yield id_, {
|
|
|
17 |
This is the XSUM subset of the GEM benchmark.
|
18 |
"""
|
19 |
_URLs = {
|
20 |
+
"xsum": {
|
21 |
+
"data": "http://bollin.inf.ed.ac.uk/public/direct/XSUM-EMNLP18-Summary-Data-Original.tar.gz",
|
22 |
+
"splits": "https://storage.googleapis.com/huggingface-nlp/datasets/gem/gem_xsum_confidence_0.8.json",
|
23 |
+
"challenge_set": "https://storage.googleapis.com/huggingface-nlp/datasets/gem/gem_challenge_sets/xsum.zip",
|
24 |
+
},
|
25 |
+
}
|
26 |
|
27 |
_XSUM_REMOVE_LINES = set(
|
28 |
[
|
|
|
40 |
]
|
41 |
)
|
42 |
|
43 |
+
|
44 |
class Xsum(datasets.GeneratorBasedBuilder):
|
45 |
|
46 |
BUILDER_CONFIGS = [
|
|
|
54 |
def _info(self):
|
55 |
return datasets.DatasetInfo(
|
56 |
description=_DESCRIPTION,
|
57 |
+
features=datasets.Features(
|
58 |
{
|
59 |
"gem_id": datasets.Value("string"),
|
60 |
"gem_parent_id": datasets.Value("string"),
|
|
|
62 |
"document": datasets.Value("string"),
|
63 |
"target": datasets.Value("string"),
|
64 |
"references": [datasets.Value("string")],
|
65 |
+
}
|
66 |
+
),
|
67 |
supervised_keys=None,
|
68 |
homepage="",
|
69 |
citation=_CITATION,
|
|
|
76 |
("challenge_train_sample", "train_xsum_RandomSample500.json"),
|
77 |
("challenge_validation_sample", "validation_xsum_RandomSample500.json"),
|
78 |
("challenge_test_backtranslation", "test_xsum_BackTranslation500.json"),
|
79 |
+
(
|
80 |
+
"challenge_test_bfp_02",
|
81 |
+
"test_xsum_ButterFingersPerturbation_p=0.02_500.json",
|
82 |
+
),
|
83 |
+
(
|
84 |
+
"challenge_test_bfp_05",
|
85 |
+
"test_xsum_ButterFingersPerturbation_p=0.05_500.json",
|
86 |
+
),
|
87 |
("challenge_test_nopunc", "test_xsum_WithoutPunctuation500.json"),
|
88 |
+
("challenge_test_covid", f"en_test_covid19.jsonl"),
|
89 |
]
|
90 |
return [
|
91 |
+
datasets.SplitGenerator(
|
92 |
+
name=datasets.Split.TRAIN,
|
93 |
+
gen_kwargs={
|
94 |
+
"filepath": dl_dir["splits"],
|
95 |
+
"split": "train",
|
96 |
+
"filepaths": os.path.join(dl_dir["data"], "bbc-summary-data"),
|
97 |
+
},
|
98 |
+
),
|
99 |
+
datasets.SplitGenerator(
|
100 |
+
name=datasets.Split.VALIDATION,
|
101 |
+
gen_kwargs={
|
102 |
+
"filepath": dl_dir["splits"],
|
103 |
+
"split": "validation",
|
104 |
+
"filepaths": os.path.join(dl_dir["data"], "bbc-summary-data"),
|
105 |
+
},
|
106 |
+
),
|
107 |
+
datasets.SplitGenerator(
|
108 |
+
name=datasets.Split.TEST,
|
109 |
+
gen_kwargs={
|
110 |
+
"filepath": dl_dir["splits"],
|
111 |
+
"split": "test",
|
112 |
+
"filepaths": os.path.join(dl_dir["data"], "bbc-summary-data"),
|
113 |
+
},
|
114 |
+
),
|
115 |
+
] + [
|
116 |
datasets.SplitGenerator(
|
117 |
name=challenge_split,
|
118 |
gen_kwargs={
|
|
|
120 |
"split": challenge_split,
|
121 |
},
|
122 |
)
|
123 |
+
for challenge_split, filename in challenge_sets
|
124 |
]
|
125 |
|
126 |
def _generate_examples(self, filepath, split, filepaths=None):
|
|
|
153 |
with open(filepath, "r", encoding="utf-8") as f:
|
154 |
split_ids = json.load(f)
|
155 |
for id_, i in enumerate(split_ids[split]):
|
156 |
+
with open(
|
157 |
+
os.path.join(filepaths, i + ".summary"), "r", encoding="utf-8"
|
158 |
+
) as f:
|
159 |
text = "".join(
|
160 |
+
[
|
161 |
+
line
|
162 |
+
for line in f.readlines()
|
163 |
+
if line not in _XSUM_REMOVE_LINES and line.strip()
|
164 |
+
]
|
165 |
)
|
166 |
segs = text.split("[SN]")
|
167 |
yield id_, {
|