Commit
•
30d835f
1
Parent(s):
d617661
update
Browse files- convert_file.sh +11 -0
- de_head_0000_2015-48.txt.gz +2 -2
- german_common_crawl.py +151 -166
- german_common_crawl.py.lock +0 -0
- write_to_filtered_file.py +25 -0
convert_file.sh
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env bash
|
| 2 |
+
file_name=${1}
|
| 3 |
+
|
| 4 |
+
local_file_name=$(echo "${file_name}" | cut -f8 -d/)
|
| 5 |
+
bare_local_file_name=$(echo "${local_file_name}" | cut -f1 -d.)
|
| 6 |
+
|
| 7 |
+
wget ${file_name}
|
| 8 |
+
|
| 9 |
+
./write_to_filtered_file.py ${local_file_name}
|
| 10 |
+
|
| 11 |
+
gzip -c "${bare_local_file_name}.txt" > "${bare_local_file_name}.txt.gz"
|
de_head_0000_2015-48.txt.gz
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f81d28cbb4c1bfaa26699540c498edaf56ca642cb74d3cd5aaad44b71ce53650
|
| 3 |
+
size 1066575032
|
german_common_crawl.py
CHANGED
|
@@ -15,10 +15,10 @@
|
|
| 15 |
"""German Common Crawl"""
|
| 16 |
|
| 17 |
from __future__ import absolute_import, division, print_function
|
| 18 |
-
|
| 19 |
-
import os
|
| 20 |
-
|
| 21 |
import datasets
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
|
| 24 |
# Find for instance the citation on arxiv or on the dataset repo/website
|
|
@@ -38,114 +38,116 @@ German Only Extract from Common Crawl
|
|
| 38 |
This Dataset is for pretraining a German Language Model (Unsupervised) or tune a Multilingual Model specifically to German
|
| 39 |
"""
|
| 40 |
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
# "
|
| 50 |
-
# "
|
| 51 |
-
# "
|
| 52 |
-
# "
|
| 53 |
-
# "
|
| 54 |
-
# "
|
| 55 |
-
# "
|
| 56 |
-
# "
|
| 57 |
-
# "
|
| 58 |
-
# "
|
| 59 |
-
# "
|
| 60 |
-
# "
|
| 61 |
-
# "
|
| 62 |
-
# "
|
| 63 |
-
# "
|
| 64 |
-
# "
|
| 65 |
-
# "
|
| 66 |
-
# "
|
| 67 |
-
# "
|
| 68 |
-
# "
|
| 69 |
-
# "
|
| 70 |
-
# "
|
| 71 |
-
# "
|
| 72 |
-
# "
|
| 73 |
-
# "
|
| 74 |
-
# "
|
| 75 |
-
# "
|
| 76 |
-
# "
|
| 77 |
-
# "
|
| 78 |
-
# "
|
| 79 |
-
# "
|
| 80 |
-
# "
|
| 81 |
-
# "
|
| 82 |
-
# "
|
| 83 |
-
# "
|
| 84 |
-
# "
|
| 85 |
-
# "
|
| 86 |
-
# "
|
| 87 |
-
# "
|
| 88 |
-
# "
|
| 89 |
-
# "
|
| 90 |
-
# "
|
| 91 |
-
# "
|
| 92 |
-
# "
|
| 93 |
-
# "
|
| 94 |
-
# "
|
| 95 |
-
# "
|
| 96 |
-
# "
|
| 97 |
-
# "
|
| 98 |
-
# "
|
| 99 |
-
# "
|
| 100 |
-
# "
|
| 101 |
-
# "
|
| 102 |
-
# "
|
| 103 |
-
# "
|
| 104 |
-
# "
|
| 105 |
-
# "
|
| 106 |
-
# "
|
| 107 |
-
# "
|
| 108 |
-
# "
|
| 109 |
-
# "
|
| 110 |
-
# "
|
| 111 |
-
# "
|
| 112 |
-
# "
|
| 113 |
-
# "
|
| 114 |
-
# "
|
| 115 |
-
# "
|
| 116 |
-
# "
|
| 117 |
-
# "
|
| 118 |
-
# "
|
| 119 |
-
# "
|
| 120 |
-
# "
|
| 121 |
-
# "
|
| 122 |
-
# "
|
| 123 |
-
# "
|
| 124 |
-
# "
|
| 125 |
-
# "
|
| 126 |
-
# "
|
| 127 |
-
# "
|
| 128 |
-
# "
|
| 129 |
-
# "
|
| 130 |
-
# "
|
| 131 |
-
# "
|
| 132 |
-
# "
|
| 133 |
-
# "
|
| 134 |
-
# "
|
| 135 |
-
# "
|
| 136 |
-
# "
|
| 137 |
-
# "
|
| 138 |
-
# "
|
| 139 |
-
# "
|
| 140 |
-
# "
|
| 141 |
-
# "
|
| 142 |
-
# "
|
| 143 |
-
# "
|
| 144 |
-
# "
|
| 145 |
-
# "
|
| 146 |
-
# "
|
| 147 |
-
# "
|
| 148 |
-
|
|
|
|
|
|
|
| 149 |
|
| 150 |
class GermanCommonCrawl(datasets.GeneratorBasedBuilder):
|
| 151 |
"""TODO: Short description of my dataset."""
|
|
@@ -161,33 +163,25 @@ class GermanCommonCrawl(datasets.GeneratorBasedBuilder):
|
|
| 161 |
]
|
| 162 |
|
| 163 |
def _info(self):
|
| 164 |
-
|
| 165 |
-
|
| 166 |
-
|
| 167 |
-
|
| 168 |
-
|
| 169 |
-
|
| 170 |
-
|
| 171 |
-
|
| 172 |
-
|
| 173 |
-
|
| 174 |
-
|
| 175 |
-
|
| 176 |
-
|
| 177 |
-
|
| 178 |
-
|
| 179 |
-
|
| 180 |
-
|
| 181 |
-
|
| 182 |
-
|
| 183 |
-
"original_nlines": datasets.Value("int32"),
|
| 184 |
-
"original_length": datasets.Value("int32"),
|
| 185 |
-
"language": datasets.Value("string"),
|
| 186 |
-
"perplexity": datasets.Value("int32"),
|
| 187 |
-
"bucket": datasets.Value("int32"),
|
| 188 |
-
|
| 189 |
-
}
|
| 190 |
-
)
|
| 191 |
return datasets.DatasetInfo(
|
| 192 |
# This is the description that will appear on the datasets page.
|
| 193 |
description=_DESCRIPTION,
|
|
@@ -203,41 +197,32 @@ class GermanCommonCrawl(datasets.GeneratorBasedBuilder):
|
|
| 203 |
|
| 204 |
def _split_generators(self, dl_manager):
|
| 205 |
"""Returns SplitGenerators."""
|
| 206 |
-
if self.config.name == "
|
| 207 |
-
|
| 208 |
else:
|
| 209 |
-
raise NotImplementedError("just `
|
| 210 |
|
| 211 |
-
import ipdb; ipdb.set_trace()
|
| 212 |
return [
|
| 213 |
datasets.SplitGenerator(
|
| 214 |
name=datasets.Split.TRAIN,
|
| 215 |
-
# These kwargs will be passed to _generate_examples
|
| 216 |
gen_kwargs={
|
| 217 |
-
"
|
| 218 |
-
"split": "train",
|
| 219 |
},
|
| 220 |
),
|
| 221 |
]
|
| 222 |
|
| 223 |
-
def _generate_examples(self,
|
| 224 |
-
"""
|
| 225 |
-
|
| 226 |
-
|
| 227 |
-
|
| 228 |
-
|
| 229 |
-
|
| 230 |
-
|
| 231 |
-
|
| 232 |
-
|
| 233 |
-
|
| 234 |
-
|
| 235 |
-
|
| 236 |
-
|
| 237 |
-
|
| 238 |
-
if self.config.name == "data_only":
|
| 239 |
-
yield id_, {
|
| 240 |
-
"raw_content": data["raw_content"],
|
| 241 |
-
}
|
| 242 |
-
else:
|
| 243 |
-
yield id_, data
|
|
|
|
| 15 |
"""German Common Crawl"""
|
| 16 |
|
| 17 |
from __future__ import absolute_import, division, print_function
|
|
|
|
|
|
|
|
|
|
| 18 |
import datasets
|
| 19 |
+
import itertools
|
| 20 |
+
import gzip
|
| 21 |
+
from ast import literal_eval
|
| 22 |
|
| 23 |
|
| 24 |
# Find for instance the citation on arxiv or on the dataset repo/website
|
|
|
|
| 38 |
This Dataset is for pretraining a German Language Model (Unsupervised) or tune a Multilingual Model specifically to German
|
| 39 |
"""
|
| 40 |
|
| 41 |
+
REPO_URL = "https://huggingface.co/datasets/flax-community/german_common_crawl/resolve/main/"
|
| 42 |
+
|
| 43 |
+
_URL_FIRST = [REPO_URL + file_name for file_name in [
|
| 44 |
+
"de_head_0000_2015-48.txt.gz",
|
| 45 |
+
]]
|
| 46 |
+
|
| 47 |
+
_URL_HEAD = [REPO_URL + file_name for file_name in [
|
| 48 |
+
"de_head_0000_2015-48.txt.gz",
|
| 49 |
+
# "de_head_0000_2016-18.txt.gz",
|
| 50 |
+
# "de_head_0000_2016-44.txt.gz",
|
| 51 |
+
# "de_head_0000_2017-13.txt.gz",
|
| 52 |
+
# "de_head_0000_2017-30.txt.gz",
|
| 53 |
+
# "de_head_0000_2017-39.txt.gz",
|
| 54 |
+
# "de_head_0000_2017-51.txt.gz",
|
| 55 |
+
# "de_head_0000_2018-09.txt.gz",
|
| 56 |
+
# "de_head_0000_2018-17.txt.gz",
|
| 57 |
+
# "de_head_0000_2018-30.txt.gz",
|
| 58 |
+
# "de_head_0000_2018-39.txt.gz",
|
| 59 |
+
# "de_head_0000_2018-51.txt.gz",
|
| 60 |
+
# "de_head_0000_2019-18.txt.gz",
|
| 61 |
+
# "de_head_0000_2019-30.txt.gz",
|
| 62 |
+
# "de_head_0000_2019-47.txt.gz",
|
| 63 |
+
# "de_head_0000_2020-10.txt.gz",
|
| 64 |
+
# "de_head_0001_2016-44.txt.gz",
|
| 65 |
+
# "de_head_0001_2017-13.txt.gz",
|
| 66 |
+
# "de_head_0001_2017-30.txt.gz",
|
| 67 |
+
# "de_head_0001_2017-39.txt.gz",
|
| 68 |
+
# "de_head_0001_2017-51.txt.gz",
|
| 69 |
+
# "de_head_0001_2018-09.txt.gz",
|
| 70 |
+
# "de_head_0001_2018-17.txt.gz",
|
| 71 |
+
# "de_head_0001_2018-30.txt.gz",
|
| 72 |
+
# "de_head_0001_2018-39.txt.gz",
|
| 73 |
+
# "de_head_0001_2018-51.txt.gz",
|
| 74 |
+
# "de_head_0001_2019-09.txt.gz",
|
| 75 |
+
# "de_head_0001_2019-18.txt.gz",
|
| 76 |
+
# "de_head_0001_2019-30.txt.gz",
|
| 77 |
+
# "de_head_0001_2019-47.txt.gz",
|
| 78 |
+
# "de_head_0001_2020-10.txt.gz",
|
| 79 |
+
# "de_head_0002_2016-44.txt.gz",
|
| 80 |
+
# "de_head_0002_2017-13.txt.gz",
|
| 81 |
+
# "de_head_0002_2017-30.txt.gz",
|
| 82 |
+
# "de_head_0002_2017-39.txt.gz",
|
| 83 |
+
# "de_head_0002_2017-51.txt.gz",
|
| 84 |
+
# "de_head_0002_2018-09.txt.gz",
|
| 85 |
+
# "de_head_0002_2018-17.txt.gz",
|
| 86 |
+
# "de_head_0002_2018-30.txt.gz",
|
| 87 |
+
# "de_head_0002_2018-39.txt.gz",
|
| 88 |
+
# "de_head_0002_2018-51.txt.gz",
|
| 89 |
+
# "de_head_0002_2019-09.txt.gz",
|
| 90 |
+
# "de_head_0002_2019-18.txt.gz",
|
| 91 |
+
# "de_head_0002_2019-30.txt.gz",
|
| 92 |
+
# "de_head_0002_2019-47.txt.gz",
|
| 93 |
+
# "de_head_0002_2020-10.txt.gz",
|
| 94 |
+
# "de_head_0003_2016-44.txt.gz",
|
| 95 |
+
# "de_head_0003_2017-13.txt.gz",
|
| 96 |
+
# "de_head_0003_2017-30.txt.gz",
|
| 97 |
+
# "de_head_0003_2017-39.txt.gz",
|
| 98 |
+
# "de_head_0003_2017-51.txt.gz",
|
| 99 |
+
# "de_head_0003_2018-09.txt.gz",
|
| 100 |
+
# "de_head_0003_2018-17.txt.gz",
|
| 101 |
+
# "de_head_0003_2018-30.txt.gz",
|
| 102 |
+
# "de_head_0003_2018-39.txt.gz",
|
| 103 |
+
# "de_head_0003_2018-51.txt.gz",
|
| 104 |
+
# "de_head_0003_2019-09.txt.gz",
|
| 105 |
+
# "de_head_0003_2019-18.txt.gz",
|
| 106 |
+
# "de_head_0003_2019-30.txt.gz",
|
| 107 |
+
# "de_head_0003_2019-47.txt.gz",
|
| 108 |
+
# "de_head_0003_2020-10.txt.gz",
|
| 109 |
+
# "de_head_0004_2016-44.txt.gz",
|
| 110 |
+
# "de_head_0004_2017-30.txt.gz",
|
| 111 |
+
# "de_head_0004_2017-39.txt.gz",
|
| 112 |
+
# "de_head_0004_2017-51.txt.gz",
|
| 113 |
+
# "de_head_0004_2018-09.txt.gz",
|
| 114 |
+
# "de_head_0004_2018-17.txt.gz",
|
| 115 |
+
# "de_head_0004_2018-30.txt.gz",
|
| 116 |
+
# "de_head_0004_2018-39.txt.gz",
|
| 117 |
+
# "de_head_0004_2018-51.txt.gz",
|
| 118 |
+
# "de_head_0004_2019-09.txt.gz",
|
| 119 |
+
# "de_head_0004_2019-18.txt.gz",
|
| 120 |
+
# "de_head_0004_2019-30.txt.gz",
|
| 121 |
+
# "de_head_0004_2019-47.txt.gz",
|
| 122 |
+
# "de_head_0004_2020-10.txt.gz",
|
| 123 |
+
# "de_head_0005_2017-51.txt.gz",
|
| 124 |
+
# "de_head_0005_2018-09.txt.gz",
|
| 125 |
+
# "de_head_0005_2018-17.txt.gz",
|
| 126 |
+
# "de_head_0005_2018-30.txt.gz",
|
| 127 |
+
# "de_head_0005_2018-39.txt.gz",
|
| 128 |
+
# "de_head_0005_2018-51.txt.gz",
|
| 129 |
+
# "de_head_0005_2019-09.txt.gz",
|
| 130 |
+
# "de_head_0005_2019-18.txt.gz",
|
| 131 |
+
# "de_head_0005_2019-30.txt.gz",
|
| 132 |
+
# "de_head_0005_2019-47.txt.gz",
|
| 133 |
+
# "de_head_0005_2020-10.txt.gz",
|
| 134 |
+
# "de_head_0006_2018-09.txt.gz",
|
| 135 |
+
# "de_head_0006_2018-17.txt.gz",
|
| 136 |
+
# "de_head_0006_2018-30.txt.gz",
|
| 137 |
+
# "de_head_0006_2018-39.txt.gz",
|
| 138 |
+
# "de_head_0006_2018-51.txt.gz",
|
| 139 |
+
# "de_head_0006_2019-09.txt.gz",
|
| 140 |
+
# "de_head_0006_2019-18.txt.gz",
|
| 141 |
+
# "de_head_0006_2019-30.txt.gz",
|
| 142 |
+
# "de_head_0006_2019-47.txt.gz",
|
| 143 |
+
# "de_head_0006_2020-10.txt.gz",
|
| 144 |
+
# "de_head_0007_2018-30.txt.gz",
|
| 145 |
+
# "de_head_0007_2018-51.txt.gz",
|
| 146 |
+
# "de_head_0007_2019-09.txt.gz",
|
| 147 |
+
# "de_head_0007_2019-18.txt.gz",
|
| 148 |
+
# "de_head_0007_2019-47.txt.gz",
|
| 149 |
+
# "de_head_0007_2020-10.txt.gz",
|
| 150 |
+
]]
|
| 151 |
|
| 152 |
class GermanCommonCrawl(datasets.GeneratorBasedBuilder):
|
| 153 |
"""TODO: Short description of my dataset."""
|
|
|
|
| 163 |
]
|
| 164 |
|
| 165 |
def _info(self):
|
| 166 |
+
features = datasets.Features(
|
| 167 |
+
{
|
| 168 |
+
"url": datasets.Value("string"),
|
| 169 |
+
"date_download": datasets.Value("string"),
|
| 170 |
+
"digest": datasets.Value("string"),
|
| 171 |
+
"length": datasets.Value("int32"),
|
| 172 |
+
"nlines": datasets.Value("int32"),
|
| 173 |
+
"source_domain": datasets.Value("string"),
|
| 174 |
+
"title": datasets.Value("string"),
|
| 175 |
+
"raw_content": datasets.Value("string"),
|
| 176 |
+
"cc_segment": datasets.Value("string"),
|
| 177 |
+
"original_nlines": datasets.Value("int32"),
|
| 178 |
+
"original_length": datasets.Value("int32"),
|
| 179 |
+
"language": datasets.Value("string"),
|
| 180 |
+
"language_score": datasets.Value("int32"),
|
| 181 |
+
"perplexity": datasets.Value("int32"),
|
| 182 |
+
"bucket": datasets.Value("string"),
|
| 183 |
+
}
|
| 184 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 185 |
return datasets.DatasetInfo(
|
| 186 |
# This is the description that will appear on the datasets page.
|
| 187 |
description=_DESCRIPTION,
|
|
|
|
| 197 |
|
| 198 |
def _split_generators(self, dl_manager):
|
| 199 |
"""Returns SplitGenerators."""
|
| 200 |
+
if self.config.name == "first":
|
| 201 |
+
data_files = dl_manager.download(_URL_FIRST)
|
| 202 |
else:
|
| 203 |
+
raise NotImplementedError("just `first` works for now")
|
| 204 |
|
|
|
|
| 205 |
return [
|
| 206 |
datasets.SplitGenerator(
|
| 207 |
name=datasets.Split.TRAIN,
|
|
|
|
| 208 |
gen_kwargs={
|
| 209 |
+
"data_files": data_files,
|
|
|
|
| 210 |
},
|
| 211 |
),
|
| 212 |
]
|
| 213 |
|
| 214 |
+
def _generate_examples(self, data_files):
|
| 215 |
+
"""This function returns the examples in the raw (text) form by iterating on all the files."""
|
| 216 |
+
for filepath in data_files:
|
| 217 |
+
with gzip.open(filepath, "rt", encoding="utf-8") as f:
|
| 218 |
+
for id_, line in enumerate(f):
|
| 219 |
+
item = literal_eval(line)
|
| 220 |
+
import ipdb; ipdb.set_trace()
|
| 221 |
+
yield id_, {
|
| 222 |
+
|
| 223 |
+
|
| 224 |
+
|
| 225 |
+
|
| 226 |
+
|
| 227 |
+
|
| 228 |
+
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
german_common_crawl.py.lock
ADDED
|
File without changes
|
write_to_filtered_file.py
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python3
|
| 2 |
+
import gzip
|
| 3 |
+
from ast import literal_eval
|
| 4 |
+
from tqdm import tqdm
|
| 5 |
+
import sys
|
| 6 |
+
|
| 7 |
+
if __name__ == '__main__':
|
| 8 |
+
file_name = sys.argv[1]
|
| 9 |
+
new_file_name = ".".join(file_name.split(".")[:1] + ["txt"])
|
| 10 |
+
with gzip.open(file_name, 'rt') as f:
|
| 11 |
+
a = f.readline()
|
| 12 |
+
a = a.split("{'url'")
|
| 13 |
+
a = [("{'url'" + item) for item in a]
|
| 14 |
+
|
| 15 |
+
b = []
|
| 16 |
+
for item in tqdm(a):
|
| 17 |
+
try:
|
| 18 |
+
if literal_eval(item)['language_score'] > 0.98:
|
| 19 |
+
b.append(item)
|
| 20 |
+
except:
|
| 21 |
+
None
|
| 22 |
+
|
| 23 |
+
with open(new_file_name, 'wt') as file_new:
|
| 24 |
+
for part in b:
|
| 25 |
+
file_new.write(part + '\n')
|