Thomas Lemberger
commited on
Commit
•
16d88ff
1
Parent(s):
2bf6656
loading script
Browse files- biolang.py +156 -0
biolang.py
ADDED
@@ -0,0 +1,156 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
|
3 |
+
#
|
4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
5 |
+
# you may not use this file except in compliance with the License.
|
6 |
+
# You may obtain a copy of the License at
|
7 |
+
#
|
8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
9 |
+
#
|
10 |
+
# Unless required by applicable law or agreed to in writing, software
|
11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
13 |
+
# See the License for the specific language governing permissions and
|
14 |
+
# limitations under the License.
|
15 |
+
|
16 |
+
|
17 |
+
# template from : https://github.com/huggingface/datasets/blob/master/templates/new_dataset_script.py
|
18 |
+
|
19 |
+
"""Loading script for the biolang dataset for language modeling in biology."""
|
20 |
+
|
21 |
+
from __future__ import absolute_import, division, print_function
|
22 |
+
|
23 |
+
import json
|
24 |
+
from pathlib import Path
|
25 |
+
import datasets
|
26 |
+
from common import CACHE
|
27 |
+
import shutil
|
28 |
+
|
29 |
+
_CITATION = """\
|
30 |
+
@Unpublished{
|
31 |
+
huggingface: dataset,
|
32 |
+
title = {biolang},
|
33 |
+
authors={Thomas Lemberger, EMBO},
|
34 |
+
year={2021}
|
35 |
+
}
|
36 |
+
"""
|
37 |
+
|
38 |
+
_DESCRIPTION = """\
|
39 |
+
This dataset is based on abstracts from the open access section of PubMed Central to train language models for the domain of biology.
|
40 |
+
"""
|
41 |
+
|
42 |
+
_HOMEPAGE = "https://europepmc.org/downloads/openaccess"
|
43 |
+
|
44 |
+
_LICENSE = "CC BY 4.0"
|
45 |
+
|
46 |
+
_URLs = {
|
47 |
+
"biolang": "https://huggingface.co/datasets/EMBO/biolang/resolve/main/oapmc_abstracts_figs.zip",
|
48 |
+
}
|
49 |
+
|
50 |
+
|
51 |
+
class BioLang(datasets.GeneratorBasedBuilder):
|
52 |
+
"""BioLang: a dataset to train language models in biology."""
|
53 |
+
|
54 |
+
VERSION = datasets.Version("0.0.1")
|
55 |
+
|
56 |
+
BUILDER_CONFIGS = [
|
57 |
+
datasets.BuilderConfig(name="MLM", version="0.0.1", description="Dataset for general masked language model."),
|
58 |
+
datasets.BuilderConfig(name="DET", version="0.0.1", description="Dataset for part-of-speech (determinant) masked language model."),
|
59 |
+
datasets.BuilderConfig(name="VERB", version="0.0.1", description="Dataset for part-of-speech (verbs) masked language model."),
|
60 |
+
datasets.BuilderConfig(name="SMALL", version="0.0.1", description="Dataset for part-of-speech (determinants, conjunctions, prepositions, pronouns) masked language model."),
|
61 |
+
]
|
62 |
+
|
63 |
+
DEFAULT_CONFIG_NAME = "MLM" # It's not mandatory to have a default configuration. Just use one if it make sense.
|
64 |
+
|
65 |
+
def _info(self):
|
66 |
+
if self.config.name == "MLM":
|
67 |
+
features = datasets.Features(
|
68 |
+
{
|
69 |
+
"input_ids": datasets.Sequence(feature=datasets.Value("int32")),
|
70 |
+
"special_tokens_mask": datasets.Sequence(feature=datasets.Value("int8")),
|
71 |
+
}
|
72 |
+
)
|
73 |
+
elif self.config.name in ["DET", "VERB", "SMALL"]:
|
74 |
+
features = datasets.Features({
|
75 |
+
"input_ids": datasets.Sequence(feature=datasets.Value("int32")),
|
76 |
+
"tag_mask": datasets.Sequence(feature=datasets.Value("int8")),
|
77 |
+
})
|
78 |
+
|
79 |
+
return datasets.DatasetInfo(
|
80 |
+
description=_DESCRIPTION,
|
81 |
+
features=features, # Here we define them above because they are different between the two configurations
|
82 |
+
supervised_keys=('input_ids', 'pos_mask'),
|
83 |
+
homepage=_HOMEPAGE,
|
84 |
+
license=_LICENSE,
|
85 |
+
citation=_CITATION,
|
86 |
+
)
|
87 |
+
|
88 |
+
def _split_generators(self, dl_manager):
|
89 |
+
"""Returns SplitGenerators."""
|
90 |
+
if self.config.data_dir:
|
91 |
+
data_dir = self.config.data_dir
|
92 |
+
else:
|
93 |
+
url = _URLs["biolang"]
|
94 |
+
data_dir = dl_manager.download_and_extract(url)
|
95 |
+
data_dir += "/oapmc_abstracts_figs"
|
96 |
+
return [
|
97 |
+
datasets.SplitGenerator(
|
98 |
+
name=datasets.Split.TRAIN,
|
99 |
+
gen_kwargs={
|
100 |
+
"filepath": data_dir + "/train.jsonl"),
|
101 |
+
"split": "train",
|
102 |
+
},
|
103 |
+
),
|
104 |
+
datasets.SplitGenerator(
|
105 |
+
name=datasets.Split.TEST,
|
106 |
+
gen_kwargs={
|
107 |
+
"filepath": data_dir + "/test.jsonl"),
|
108 |
+
"split": "test"
|
109 |
+
},
|
110 |
+
),
|
111 |
+
datasets.SplitGenerator(
|
112 |
+
name=datasets.Split.VALIDATION,
|
113 |
+
gen_kwargs={
|
114 |
+
"filepath": data_dir + "/eval.jsonl"),
|
115 |
+
"split": "eval",
|
116 |
+
},
|
117 |
+
),
|
118 |
+
]
|
119 |
+
|
120 |
+
def _generate_examples(self, filepath, split):
|
121 |
+
""" Yields examples. """
|
122 |
+
with open(filepath, encoding="utf-8") as f:
|
123 |
+
for id_, row in enumerate(f):
|
124 |
+
data = json.loads(row)
|
125 |
+
if self.config.name == "MLM":
|
126 |
+
yield id_, {
|
127 |
+
"input_ids": data["input_ids"],
|
128 |
+
"special_tokens_mask": data['special_tokens_mask']
|
129 |
+
}
|
130 |
+
elif self.config.name == "DET":
|
131 |
+
pos_mask = [0] * len(data['input_ids'])
|
132 |
+
for idx, label in enumerate(data['label_ids']):
|
133 |
+
if label == 'DET':
|
134 |
+
pos_mask[idx] = 1
|
135 |
+
yield id_, {
|
136 |
+
"input_ids": data['input_ids'],
|
137 |
+
"tag_mask": pos_mask,
|
138 |
+
}
|
139 |
+
elif self.config.name == "VERB":
|
140 |
+
pos_mask = [0] * len(data['input_ids'])
|
141 |
+
for idx, label in enumerate(data['label_ids']):
|
142 |
+
if label == 'VERB':
|
143 |
+
pos_mask[idx] = 1
|
144 |
+
yield id_, {
|
145 |
+
"input_ids": data['input_ids'],
|
146 |
+
"tag_mask": pos_mask,
|
147 |
+
}
|
148 |
+
elif self.config.name == "SMALL":
|
149 |
+
pos_mask = [0] * len(data['input_ids'])
|
150 |
+
for idx, label in enumerate(data['label_ids']):
|
151 |
+
if label in ['DET', 'CCONJ', 'SCONJ', 'ADP', 'PRON']:
|
152 |
+
pos_mask[idx] = 1
|
153 |
+
yield id_, {
|
154 |
+
"input_ids": data['input_ids'],
|
155 |
+
"tag_mask": pos_mask,
|
156 |
+
}
|