Maurice Weber
commited on
Commit
•
7f1cb5c
1
Parent(s):
19f2e7e
subsamples 10B, 100B, 1T
Browse files- .gitignore +32 -0
- RedPajama-Data-V2.py +93 -59
.gitignore
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Byte-compiled / optimized / DLL files
|
2 |
+
__pycache__/
|
3 |
+
*.py[cod]
|
4 |
+
*$py.class
|
5 |
+
*.pyc
|
6 |
+
*.DS_Store
|
7 |
+
|
8 |
+
# data folders
|
9 |
+
data/*
|
10 |
+
!data/.gitkeep
|
11 |
+
|
12 |
+
# notebooks
|
13 |
+
notebooks/*
|
14 |
+
.ipynb_checkpoints
|
15 |
+
|
16 |
+
# Environments
|
17 |
+
.env
|
18 |
+
.venv
|
19 |
+
env/
|
20 |
+
venv/
|
21 |
+
ENV/
|
22 |
+
env.bak/
|
23 |
+
venv.bak/
|
24 |
+
|
25 |
+
# ides
|
26 |
+
.idea/
|
27 |
+
.vscode/
|
28 |
+
|
29 |
+
# distribution
|
30 |
+
*.egg-info/
|
31 |
+
dist/
|
32 |
+
build/
|
RedPajama-Data-V2.py
CHANGED
@@ -15,13 +15,12 @@
|
|
15 |
# Lint as: python3
|
16 |
"""RedPajama V2: Quality annotated Web Text Documents."""
|
17 |
|
|
|
18 |
import json
|
19 |
-
|
20 |
-
import datasets
|
21 |
import traceback
|
22 |
-
import gzip
|
23 |
from typing import List
|
24 |
|
|
|
25 |
import pyarrow.parquet as pq
|
26 |
|
27 |
logger = datasets.logging.get_logger(__name__)
|
@@ -30,10 +29,11 @@ _DESCRIPTION = """\
|
|
30 |
RedPajama V2: an Open Dataset for Training Large Language Models
|
31 |
"""
|
32 |
|
33 |
-
_URL_BASE =
|
34 |
_LANGUAGES = ("en", "de", "fr", "es", "it")
|
35 |
_MISSING_FILES_PATTERN = "urls/missing-{component}.txt"
|
36 |
_NUM_SHARDS = 5000
|
|
|
37 |
|
38 |
_CC_SNAPSHOT_IDS = (
|
39 |
"2014-15",
|
@@ -119,7 +119,7 @@ _CC_SNAPSHOT_IDS = (
|
|
119 |
"2022-40",
|
120 |
"2022-49",
|
121 |
"2023-06",
|
122 |
-
"2023-14"
|
123 |
)
|
124 |
|
125 |
|
@@ -138,19 +138,34 @@ class RedPajamaDataV2Config(datasets.BuilderConfig):
|
|
138 |
|
139 |
|
140 |
class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
141 |
-
"""
|
142 |
|
143 |
BUILDER_CONFIGS = [
|
144 |
RedPajamaDataV2Config(
|
145 |
-
name=
|
146 |
version=datasets.Version("1.0.0", ""),
|
147 |
description=f"RedPajamaV2 Sample",
|
148 |
),
|
149 |
RedPajamaDataV2Config(
|
150 |
-
name=
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
151 |
version=datasets.Version("1.0.0", ""),
|
152 |
description=f"RedPajamaV2",
|
153 |
-
)
|
154 |
]
|
155 |
|
156 |
def _info(self):
|
@@ -161,7 +176,7 @@ class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
|
161 |
"raw_content": datasets.Value("string"),
|
162 |
"doc_id": datasets.Value("string"),
|
163 |
"meta": datasets.Value("string"),
|
164 |
-
"quality_signals": datasets.Value("string")
|
165 |
}
|
166 |
),
|
167 |
supervised_keys=None,
|
@@ -177,24 +192,30 @@ class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
|
177 |
|
178 |
# fetch documents
|
179 |
logger.info(f"Downloading {len(sample_base_tags)} documents files.")
|
180 |
-
documents_files = dl_manager.download(
|
181 |
-
|
182 |
-
|
183 |
-
|
|
|
|
|
184 |
|
185 |
# fetch quality signals
|
186 |
logger.info(f"Downloading {len(sample_base_tags)} quality signals files.")
|
187 |
-
quality_signals_files = dl_manager.download(
|
188 |
-
|
189 |
-
|
190 |
-
|
|
|
|
|
191 |
|
192 |
# fetch ids of duplicates
|
193 |
logger.info(f"Downloading {len(sample_base_tags)} duplicates ids files.")
|
194 |
-
duplicates_ids_files = dl_manager.download(
|
195 |
-
|
196 |
-
|
197 |
-
|
|
|
|
|
198 |
|
199 |
return [
|
200 |
datasets.SplitGenerator(
|
@@ -203,31 +224,41 @@ class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
|
203 |
"base_tags": sample_base_tags,
|
204 |
"documents_files": documents_files,
|
205 |
"quality_signals_files": quality_signals_files,
|
206 |
-
"duplicates_ids_files": duplicates_ids_files
|
207 |
-
}
|
208 |
)
|
209 |
]
|
210 |
|
211 |
def _split_generators_full(self, dl_manager):
|
212 |
-
snapshots = getattr(self.config,
|
213 |
-
languages = getattr(self.config,
|
214 |
-
partition = getattr(self.config,
|
215 |
-
|
216 |
-
if
|
217 |
-
|
218 |
-
|
219 |
-
|
220 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
221 |
partitions = [partition]
|
222 |
else:
|
223 |
-
raise ValueError(f
|
224 |
|
225 |
# fetch list of missing files (e.g., missing duplicates or corrupted documents and
|
226 |
# quality signal files)
|
227 |
-
missing_files_paths = dl_manager.download_and_extract(
|
228 |
-
|
229 |
-
|
230 |
-
|
|
|
|
|
231 |
|
232 |
missing_files = {}
|
233 |
for component, missing_file in missing_files_paths.items():
|
@@ -243,7 +274,7 @@ class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
|
243 |
for lang in languages:
|
244 |
for snapshot in snapshots:
|
245 |
for part in partitions:
|
246 |
-
for n in range(
|
247 |
base_tag = f"{snapshot}/{n:04d}/{lang}_{part}"
|
248 |
base_tags.append(base_tag)
|
249 |
|
@@ -281,20 +312,19 @@ class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
|
281 |
"base_tags": base_tags,
|
282 |
"documents_files": documents_files,
|
283 |
"quality_signals_files": quality_signals_files,
|
284 |
-
"duplicates_ids_files": duplicates_ids_files
|
285 |
-
}
|
286 |
)
|
287 |
]
|
288 |
|
289 |
def _split_generators(self, dl_manager):
|
290 |
-
if self.config.name
|
291 |
return self._split_generators_sample(dl_manager)
|
292 |
|
293 |
return self._split_generators_full(dl_manager)
|
294 |
|
295 |
def _generate_examples(
|
296 |
-
|
297 |
-
duplicates_ids_files
|
298 |
):
|
299 |
key = 0
|
300 |
for base_tag in base_tags:
|
@@ -305,9 +335,7 @@ class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
|
305 |
if doc_file is None:
|
306 |
continue
|
307 |
|
308 |
-
for sample in self.__get_generator(
|
309 |
-
base_tag, doc_file, qs_file, dupe_file
|
310 |
-
):
|
311 |
yield key, sample
|
312 |
key += 1
|
313 |
|
@@ -325,19 +353,17 @@ class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
|
325 |
try:
|
326 |
yield self.handle_record("tail", doc_id, doc, None, None)
|
327 |
except Exception as e:
|
328 |
-
logger.warning(f
|
329 |
traceback.print_exc()
|
330 |
continue
|
331 |
|
332 |
except gzip.BadGzipFile as e:
|
333 |
# skip broken gzip files
|
334 |
-
print(f
|
335 |
traceback.print_exc()
|
336 |
return
|
337 |
|
338 |
-
def _handle_head_middle(
|
339 |
-
self, base_tag, doc_file, qs_file, dupe_file
|
340 |
-
):
|
341 |
if qs_file is None:
|
342 |
yield from self._handle_tail(base_tag, doc_file, None, None)
|
343 |
return
|
@@ -345,11 +371,13 @@ class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
|
345 |
# load duplicates
|
346 |
try:
|
347 |
with open(dupe_file, "rb") as df:
|
348 |
-
duplicates = set(
|
349 |
-
df, columns=["doc_id"], use_pandas_metadata=False
|
350 |
-
|
|
|
|
|
351 |
except Exception as e:
|
352 |
-
logger.warning(f
|
353 |
duplicates = set()
|
354 |
|
355 |
try:
|
@@ -360,16 +388,22 @@ class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
|
360 |
|
361 |
try:
|
362 |
yield self.handle_record(
|
363 |
-
"head_middle",
|
|
|
|
|
|
|
|
|
364 |
)
|
365 |
except Exception as e:
|
366 |
-
logger.warning(
|
|
|
|
|
367 |
traceback.print_exc()
|
368 |
continue
|
369 |
|
370 |
except gzip.BadGzipFile as e:
|
371 |
# skip broken gzip files
|
372 |
-
print(f
|
373 |
traceback.print_exc()
|
374 |
return
|
375 |
|
|
|
15 |
# Lint as: python3
|
16 |
"""RedPajama V2: Quality annotated Web Text Documents."""
|
17 |
|
18 |
+
import gzip
|
19 |
import json
|
|
|
|
|
20 |
import traceback
|
|
|
21 |
from typing import List
|
22 |
|
23 |
+
import datasets
|
24 |
import pyarrow.parquet as pq
|
25 |
|
26 |
logger = datasets.logging.get_logger(__name__)
|
|
|
29 |
RedPajama V2: an Open Dataset for Training Large Language Models
|
30 |
"""
|
31 |
|
32 |
+
_URL_BASE = "https://data.together.xyz/redpajama-data-v2/v1.0.0"
|
33 |
_LANGUAGES = ("en", "de", "fr", "es", "it")
|
34 |
_MISSING_FILES_PATTERN = "urls/missing-{component}.txt"
|
35 |
_NUM_SHARDS = 5000
|
36 |
+
_SUBSAMPLE_FILE_COUNTS = {"sample-10B": 1, "sample-100B": 10, "sample-1T": 100}
|
37 |
|
38 |
_CC_SNAPSHOT_IDS = (
|
39 |
"2014-15",
|
|
|
119 |
"2022-40",
|
120 |
"2022-49",
|
121 |
"2023-06",
|
122 |
+
"2023-14",
|
123 |
)
|
124 |
|
125 |
|
|
|
138 |
|
139 |
|
140 |
class RedPajamaV2(datasets.GeneratorBasedBuilder):
|
141 |
+
"""RedPajama V2: Quality annotated Web Text Documents."""
|
142 |
|
143 |
BUILDER_CONFIGS = [
|
144 |
RedPajamaDataV2Config(
|
145 |
+
name="sample",
|
146 |
version=datasets.Version("1.0.0", ""),
|
147 |
description=f"RedPajamaV2 Sample",
|
148 |
),
|
149 |
RedPajamaDataV2Config(
|
150 |
+
name="sample-10B",
|
151 |
+
version=datasets.Version("1.0.0", ""),
|
152 |
+
description=f"RedPajamaV2 Sample with 10B tokens",
|
153 |
+
),
|
154 |
+
RedPajamaDataV2Config(
|
155 |
+
name="sample-100B",
|
156 |
+
version=datasets.Version("1.0.0", ""),
|
157 |
+
description=f"RedPajamaV2 Sample with 100B tokens",
|
158 |
+
),
|
159 |
+
RedPajamaDataV2Config(
|
160 |
+
name="sample-1T",
|
161 |
+
version=datasets.Version("1.0.0", ""),
|
162 |
+
description=f"RedPajamaV2 Sample with 1T tokens",
|
163 |
+
),
|
164 |
+
RedPajamaDataV2Config(
|
165 |
+
name="default",
|
166 |
version=datasets.Version("1.0.0", ""),
|
167 |
description=f"RedPajamaV2",
|
168 |
+
),
|
169 |
]
|
170 |
|
171 |
def _info(self):
|
|
|
176 |
"raw_content": datasets.Value("string"),
|
177 |
"doc_id": datasets.Value("string"),
|
178 |
"meta": datasets.Value("string"),
|
179 |
+
"quality_signals": datasets.Value("string"),
|
180 |
}
|
181 |
),
|
182 |
supervised_keys=None,
|
|
|
192 |
|
193 |
# fetch documents
|
194 |
logger.info(f"Downloading {len(sample_base_tags)} documents files.")
|
195 |
+
documents_files = dl_manager.download(
|
196 |
+
{
|
197 |
+
base_tag: f"sample/documents/{base_tag}.json.gz"
|
198 |
+
for base_tag in sample_base_tags
|
199 |
+
}
|
200 |
+
)
|
201 |
|
202 |
# fetch quality signals
|
203 |
logger.info(f"Downloading {len(sample_base_tags)} quality signals files.")
|
204 |
+
quality_signals_files = dl_manager.download(
|
205 |
+
{
|
206 |
+
base_tag: f"sample/quality_signals/{base_tag}.signals.json.gz"
|
207 |
+
for base_tag in sample_base_tags
|
208 |
+
}
|
209 |
+
)
|
210 |
|
211 |
# fetch ids of duplicates
|
212 |
logger.info(f"Downloading {len(sample_base_tags)} duplicates ids files.")
|
213 |
+
duplicates_ids_files = dl_manager.download(
|
214 |
+
{
|
215 |
+
base_tag: f"sample/duplicates/{base_tag}.duplicates.parquet"
|
216 |
+
for base_tag in sample_base_tags
|
217 |
+
}
|
218 |
+
)
|
219 |
|
220 |
return [
|
221 |
datasets.SplitGenerator(
|
|
|
224 |
"base_tags": sample_base_tags,
|
225 |
"documents_files": documents_files,
|
226 |
"quality_signals_files": quality_signals_files,
|
227 |
+
"duplicates_ids_files": duplicates_ids_files,
|
228 |
+
},
|
229 |
)
|
230 |
]
|
231 |
|
232 |
def _split_generators_full(self, dl_manager):
|
233 |
+
snapshots = getattr(self.config, "snapshots", _CC_SNAPSHOT_IDS)
|
234 |
+
languages = getattr(self.config, "languages", _LANGUAGES)
|
235 |
+
partition = getattr(self.config, "partition", "all")
|
236 |
+
|
237 |
+
if self.config.name in ("sample-10B", "sample-100B", "sample-1T"):
|
238 |
+
partition = "head_middle"
|
239 |
+
languages = _LANGUAGES
|
240 |
+
snapshots = _CC_SNAPSHOT_IDS
|
241 |
+
num_shards = _SUBSAMPLE_FILE_COUNTS[self.config.name]
|
242 |
+
else:
|
243 |
+
num_shards = _NUM_SHARDS
|
244 |
+
|
245 |
+
if partition == "all":
|
246 |
+
partitions = ["head", "middle", "tail"]
|
247 |
+
elif partition == "head_middle":
|
248 |
+
partitions = ["head", "middle"]
|
249 |
+
elif partition == "tail":
|
250 |
partitions = [partition]
|
251 |
else:
|
252 |
+
raise ValueError(f"invalid partition: {partition}")
|
253 |
|
254 |
# fetch list of missing files (e.g., missing duplicates or corrupted documents and
|
255 |
# quality signal files)
|
256 |
+
missing_files_paths = dl_manager.download_and_extract(
|
257 |
+
{
|
258 |
+
component: _MISSING_FILES_PATTERN.format(component=component)
|
259 |
+
for component in ("documents", "signals", "duplicates")
|
260 |
+
}
|
261 |
+
)
|
262 |
|
263 |
missing_files = {}
|
264 |
for component, missing_file in missing_files_paths.items():
|
|
|
274 |
for lang in languages:
|
275 |
for snapshot in snapshots:
|
276 |
for part in partitions:
|
277 |
+
for n in range(num_shards):
|
278 |
base_tag = f"{snapshot}/{n:04d}/{lang}_{part}"
|
279 |
base_tags.append(base_tag)
|
280 |
|
|
|
312 |
"base_tags": base_tags,
|
313 |
"documents_files": documents_files,
|
314 |
"quality_signals_files": quality_signals_files,
|
315 |
+
"duplicates_ids_files": duplicates_ids_files,
|
316 |
+
},
|
317 |
)
|
318 |
]
|
319 |
|
320 |
def _split_generators(self, dl_manager):
|
321 |
+
if self.config.name == "sample":
|
322 |
return self._split_generators_sample(dl_manager)
|
323 |
|
324 |
return self._split_generators_full(dl_manager)
|
325 |
|
326 |
def _generate_examples(
|
327 |
+
self, base_tags, documents_files, quality_signals_files, duplicates_ids_files
|
|
|
328 |
):
|
329 |
key = 0
|
330 |
for base_tag in base_tags:
|
|
|
335 |
if doc_file is None:
|
336 |
continue
|
337 |
|
338 |
+
for sample in self.__get_generator(base_tag, doc_file, qs_file, dupe_file):
|
|
|
|
|
339 |
yield key, sample
|
340 |
key += 1
|
341 |
|
|
|
353 |
try:
|
354 |
yield self.handle_record("tail", doc_id, doc, None, None)
|
355 |
except Exception as e:
|
356 |
+
logger.warning(f"failed handling row {row} in {doc_file}")
|
357 |
traceback.print_exc()
|
358 |
continue
|
359 |
|
360 |
except gzip.BadGzipFile as e:
|
361 |
# skip broken gzip files
|
362 |
+
print(f"BadGzipFile: {doc_file, qs_file}")
|
363 |
traceback.print_exc()
|
364 |
return
|
365 |
|
366 |
+
def _handle_head_middle(self, base_tag, doc_file, qs_file, dupe_file):
|
|
|
|
|
367 |
if qs_file is None:
|
368 |
yield from self._handle_tail(base_tag, doc_file, None, None)
|
369 |
return
|
|
|
371 |
# load duplicates
|
372 |
try:
|
373 |
with open(dupe_file, "rb") as df:
|
374 |
+
duplicates = set(
|
375 |
+
pq.read_table(df, columns=["doc_id"], use_pandas_metadata=False)[
|
376 |
+
"doc_id"
|
377 |
+
].to_pylist()
|
378 |
+
)
|
379 |
except Exception as e:
|
380 |
+
logger.warning(f"no duplicate ids found for {base_tag}")
|
381 |
duplicates = set()
|
382 |
|
383 |
try:
|
|
|
388 |
|
389 |
try:
|
390 |
yield self.handle_record(
|
391 |
+
part="head_middle",
|
392 |
+
doc_id=doc_id,
|
393 |
+
doc=doc,
|
394 |
+
qs=qs,
|
395 |
+
is_duplicate=doc_id in duplicates,
|
396 |
)
|
397 |
except Exception as e:
|
398 |
+
logger.warning(
|
399 |
+
f"failed handling row {row} in {doc_file} ({qs_file})"
|
400 |
+
)
|
401 |
traceback.print_exc()
|
402 |
continue
|
403 |
|
404 |
except gzip.BadGzipFile as e:
|
405 |
# skip broken gzip files
|
406 |
+
print(f"BadGzipFile: {doc_file, qs_file}")
|
407 |
traceback.print_exc()
|
408 |
return
|
409 |
|