Add v20220616 release files
Browse files- .gitattributes +2 -0
- multi_lexsum.py +155 -0
- releases/v20220616/dev.json +3 -0
- releases/v20220616/sources.json +3 -0
- releases/v20220616/test.json +3 -0
- releases/v20220616/train.json +3 -0
.gitattributes
CHANGED
@@ -39,3 +39,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
39 |
*.mp3 filter=lfs diff=lfs merge=lfs -text
|
40 |
*.ogg filter=lfs diff=lfs merge=lfs -text
|
41 |
*.wav filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
39 |
*.mp3 filter=lfs diff=lfs merge=lfs -text
|
40 |
*.ogg filter=lfs diff=lfs merge=lfs -text
|
41 |
*.wav filter=lfs diff=lfs merge=lfs -text
|
42 |
+
# MultiLexSum Specific
|
43 |
+
*.json filter=lfs diff=lfs merge=lfs -text
|
multi_lexsum.py
ADDED
@@ -0,0 +1,155 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from typing import List, Union, Dict, Any, Tuple
|
2 |
+
import json
|
3 |
+
import os
|
4 |
+
|
5 |
+
import datasets
|
6 |
+
from datasets.tasks import Summarization
|
7 |
+
|
8 |
+
logger = datasets.logging.get_logger(__name__)
|
9 |
+
|
10 |
+
|
11 |
+
def _load_jsonl(filename):
|
12 |
+
with open(filename, "r") as fp:
|
13 |
+
jsonl_content = fp.read()
|
14 |
+
|
15 |
+
result = [json.loads(jline) for jline in jsonl_content.splitlines()]
|
16 |
+
return result
|
17 |
+
|
18 |
+
|
19 |
+
def _load_json(filepath):
|
20 |
+
|
21 |
+
with open(filepath, "r") as fp:
|
22 |
+
res = json.load(fp)
|
23 |
+
return res
|
24 |
+
|
25 |
+
|
26 |
+
_CITATION = """
|
27 |
+
@article{Shen2022MultiLexSum,
|
28 |
+
author = {Zejiang Shen and
|
29 |
+
Kyle Lo and
|
30 |
+
Lauren Yu and
|
31 |
+
Nathan Dahlberg and
|
32 |
+
Margo Schlanger and
|
33 |
+
Doug Downey},
|
34 |
+
title = {Multi-LexSum: Real-World Summaries of Civil Rights Lawsuits at Multiple Granularities},
|
35 |
+
journal = {CoRR},
|
36 |
+
volume = {abs/2206.10883},
|
37 |
+
year = {2022},
|
38 |
+
url = {https://doi.org/10.48550/arXiv.2206.10883},
|
39 |
+
doi = {10.48550/arXiv.2206.10883}
|
40 |
+
}
|
41 |
+
""" # TODO
|
42 |
+
|
43 |
+
_DESCRIPTION = """
|
44 |
+
Multi-LexSum is a multi-doc summarization dataset for civil rights litigation lawsuits with summaries of three granularities.
|
45 |
+
""" # TODO: Update with full abstract
|
46 |
+
|
47 |
+
_HOMEPAGE = "https://multilexsum.github.io"
|
48 |
+
|
49 |
+
# _BASE_URL = "https://ai2-s2-research.s3.us-west-2.amazonaws.com/multilexsum/releases"
|
50 |
+
_BASE_URL = "https://huggingface.co/datasets/allenai/multi_lexsum/resolve/main/releases"
|
51 |
+
_FILES = {
|
52 |
+
"train": "train.json",
|
53 |
+
"dev": "dev.json",
|
54 |
+
"test": "test.json",
|
55 |
+
"sources": "sources.json",
|
56 |
+
}
|
57 |
+
|
58 |
+
|
59 |
+
class MultiLexsumConfig(datasets.BuilderConfig):
|
60 |
+
"""BuilderConfig for LexSum."""
|
61 |
+
|
62 |
+
def __init__(self, **kwargs):
|
63 |
+
"""BuilderConfig for LexSum.
|
64 |
+
Args:
|
65 |
+
**kwargs: keyword arguments forwarded to super.
|
66 |
+
"""
|
67 |
+
super(MultiLexsumConfig, self).__init__(**kwargs)
|
68 |
+
|
69 |
+
|
70 |
+
class MultiLexsum(datasets.GeneratorBasedBuilder):
|
71 |
+
"""MultiLexSum Dataset: a multi-doc summarization dataset for
|
72 |
+
civil rights litigation lawsuits with summaries of three granularities.
|
73 |
+
"""
|
74 |
+
|
75 |
+
BUILDER_CONFIGS = [
|
76 |
+
MultiLexsumConfig(
|
77 |
+
name="v20220616",
|
78 |
+
version=datasets.Version("1.0.0", "Public v1.0 release."),
|
79 |
+
description="The v1.0 Multi-LexSum dataset",
|
80 |
+
),
|
81 |
+
]
|
82 |
+
|
83 |
+
def _info(self):
|
84 |
+
return datasets.DatasetInfo(
|
85 |
+
description=_DESCRIPTION,
|
86 |
+
features=datasets.Features(
|
87 |
+
{
|
88 |
+
"id": datasets.Value("string"),
|
89 |
+
"sources": datasets.Sequence(datasets.Value("string")),
|
90 |
+
"summary/long": datasets.Value("string"),
|
91 |
+
"summary/short": datasets.Value("string"),
|
92 |
+
"summary/tiny": datasets.Value("string"),
|
93 |
+
}
|
94 |
+
),
|
95 |
+
supervised_keys=None,
|
96 |
+
homepage=_HOMEPAGE,
|
97 |
+
citation=_CITATION,
|
98 |
+
task_templates=[
|
99 |
+
Summarization(text_column="source", summary_column="summary/long")
|
100 |
+
],
|
101 |
+
)
|
102 |
+
|
103 |
+
def _split_generators(self, dl_manager):
|
104 |
+
|
105 |
+
base_url = _BASE_URL if self.config.data_dir is None else self.config.data_dir
|
106 |
+
downloaded_files = dl_manager.download_and_extract(
|
107 |
+
{
|
108 |
+
name: f"{base_url}/{self.config.name}/{filename}"
|
109 |
+
for name, filename in _FILES.items()
|
110 |
+
}
|
111 |
+
)
|
112 |
+
# Given sources is a large file, we read it first
|
113 |
+
sources = _load_json(downloaded_files["sources"])
|
114 |
+
|
115 |
+
return [
|
116 |
+
datasets.SplitGenerator(
|
117 |
+
name=datasets.Split.TRAIN,
|
118 |
+
gen_kwargs={
|
119 |
+
"subset_file": downloaded_files["train"],
|
120 |
+
"sources": sources,
|
121 |
+
},
|
122 |
+
),
|
123 |
+
datasets.SplitGenerator(
|
124 |
+
name=datasets.Split.VALIDATION,
|
125 |
+
gen_kwargs={
|
126 |
+
"subset_file": downloaded_files["dev"],
|
127 |
+
"sources": sources,
|
128 |
+
},
|
129 |
+
),
|
130 |
+
datasets.SplitGenerator(
|
131 |
+
name=datasets.Split.TEST,
|
132 |
+
gen_kwargs={
|
133 |
+
"subset_file": downloaded_files["test"],
|
134 |
+
"sources": sources,
|
135 |
+
},
|
136 |
+
),
|
137 |
+
]
|
138 |
+
|
139 |
+
def _generate_examples(self, subset_file: str, sources: Dict[str, Dict]):
|
140 |
+
"""This function returns the examples in the raw (text) form."""
|
141 |
+
logger.info(f"generating examples from = {subset_file}")
|
142 |
+
|
143 |
+
subset_cases = _load_jsonl(subset_file)
|
144 |
+
for case_data in subset_cases:
|
145 |
+
case_sources = [
|
146 |
+
sources[source_id]["doc_text"]
|
147 |
+
for source_id in case_data["case_documents"]
|
148 |
+
]
|
149 |
+
yield case_data["case_id"], {
|
150 |
+
"id": case_data["case_id"],
|
151 |
+
"sources": case_sources,
|
152 |
+
"summary/long": case_data["summary/long"],
|
153 |
+
"summary/short": case_data["summary/short"],
|
154 |
+
"summary/tiny": case_data["summary/tiny"],
|
155 |
+
}
|
releases/v20220616/dev.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:00b5110633b1b3d5c33e6cc4645525b3633cbb1844fcdc63315f6b2b340fa958
|
3 |
+
size 2281645
|
releases/v20220616/sources.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d15a29d05ee4c8052270bf630633ec5b8c5ff7dab6cc480c009dc50f76c8ce24
|
3 |
+
size 2219115572
|
releases/v20220616/test.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:64fafb058f84cef09fc3d89df3ef913ee824d28f62bde9c1c9f52cc4d7c5b40a
|
3 |
+
size 4272330
|
releases/v20220616/train.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:edb45aee04a4aa1eebce2ec05880304322f29baf1b7bcf2c23409c020cae6aca
|
3 |
+
size 15711733
|