Make the dataset streamable

.gitattributes

@@ -25,3 +25,24 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zstandard filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+urlsf_subset00.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset01.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset02.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset03.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset04.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset05.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset06.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset07.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset08.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset09.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset10.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset11.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset12.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset13.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset14.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset15.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset16.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset17.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset18.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset19.tar filter=lfs diff=lfs merge=lfs -text
+urlsf_subset20.tar filter=lfs diff=lfs merge=lfs -text

README.md

@@ -28,10 +28,10 @@ dataset_info:
   config_name: plain_text
   splits:
   - name: train
-    num_bytes: 39769494896
+    num_bytes: 39769491688
     num_examples: 8013769
-  download_size: 12880027468
-  dataset_size: 39769494896
+  download_size: 12880189440
+  dataset_size: 39769491688
 ---
 
 # Dataset Card for "openwebtext"
@@ -72,7 +72,9 @@ dataset_info:
 
 ### Dataset Summary
 
-An open-source replication of the WebText dataset from OpenAI.
+An open-source replication of the WebText dataset from OpenAI, that was used to train GPT-2.
+
+This distribution was created by Aaron Gokaslan and Vanya Cohen of Brown University.
 
 ### Supported Tasks and Leaderboards
 
@@ -124,7 +126,9 @@ The data fields are the same among all splits.
 
 #### Initial Data Collection and Normalization
 
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
+The authors started by extracting all Reddit post urls from the Reddit submissions dataset. These links were deduplicated, filtered to exclude non-html content, and then shuffled randomly. The links were then distributed to several machines in parallel for download, and all web pages were extracted using the newspaper python package. Using Facebook FastText, non-English web pages were filtered out.
+
+Subsequently, near-duplicate documents were identified using local-sensitivity hashing (LSH). Documents were hashed into sets of 5-grams and all documents that had a similarity threshold of greater than 0.5 were removed. The the remaining documents were tokenized, and documents with fewer than 128 tokens were removed. This left 38GB of text data (40GB using SI units) from 8,013,769 documents.
 
 #### Who are the source language producers?
 
@@ -132,13 +136,7 @@ The data fields are the same among all splits.
 
 ### Annotations
 
-#### Annotation process
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
-
-#### Who are the annotators?
-
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
+The dataset doesn't contain annotations.
 
 ### Personal and Sensitive Information
 
@@ -166,7 +164,30 @@ The data fields are the same among all splits.
 
 ### Licensing Information
 
-[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
+These data are released under this licensing scheme from the original authors ([source](https://skylion007.github.io/OpenWebTextCorpus/)):
+
+```
+We do not own any of the text from which these data has been extracted.
+
+We license the actual packaging of these parallel data under the [Creative Commons CC0 license (“no rights reserved”)](https://creativecommons.org/share-your-work/public-domain/cc0/)
+```
+
+#### Notice policy
+
+Should you consider that our data contains material that is owned by you and should therefore not be reproduced here, please:
+
+Clearly identify yourself, with detailed contact data such as an address, telephone number or email address at which you can be contacted.
+
+Clearly identify the copyrighted work claimed to be infringed.
+
+Clearly identify the material that is claimed to be infringing and information reasonably sufficient to allow us to locate the material.
+
+And contact us at the following email address: openwebtext at gmail.com and datasets at huggingface.co
+
+#### Take down policy
+
+The original authors will comply to legitimate requests by removing the affected sources from the next release of the corpus.
+Hugging Face will also update this repository accordingly.
 
 ### Citation Information
 
@@ -181,4 +202,4 @@ The data fields are the same among all splits.
 
 ### Contributions
 
-Thanks to [@richarddwang](https://github.com/richarddwang) for adding this dataset.
+Thanks to [@richarddwang](https://github.com/richarddwang) for adding this dataset.

dataset_infos.json

@@ -1 +0,0 @@
-{"plain_text": {"description": "An open-source replication of the WebText dataset from OpenAI.\n", "citation": "@misc{Gokaslan2019OpenWeb,\n  title={OpenWebText Corpus},\n  author={Aaron Gokaslan*, Vanya Cohen*, Ellie Pavlick, Stefanie Tellex},\n  howpublished{\\url{http://Skylion007.github.io/OpenWebTextCorpus}},\n  year={2019}\n}\n", "homepage": "https://skylion007.github.io/OpenWebTextCorpus/", "license": "", "features": {"text": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "openwebtext", "config_name": "plain_text", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 39769494896, "num_examples": 8013769, "dataset_name": "openwebtext"}}, "download_checksums": {"https://zenodo.org/record/3834942/files/openwebtext.tar.xz": {"num_bytes": 12880027468, "checksum": "9fe39d154c5bc67da8c359415372b79510eb1e2edb0d035fe4f7fc3a732b9336"}}, "download_size": 12880027468, "post_processing_size": null, "dataset_size": 39769494896, "size_in_bytes": 52649522364}}

openwebtext.py

@@ -14,10 +14,7 @@
 # limitations under the License.
 """The Open WebText Corpus"""
 
-
-import os
 import re
-from itertools import chain
 
 import datasets
 
@@ -35,7 +32,8 @@ _DESCRIPTION = """\
 An open-source replication of the WebText dataset from OpenAI.
 """
 
-_URL = "https://zenodo.org/record/3834942/files/openwebtext.tar.xz"
+_N_DATA_FILES = 21
+_DATA_FILES = ["subsets/urlsf_subset{:02d}.tar".format(i) for i in range(_N_DATA_FILES)]
 
 
 class Openwebtext(datasets.GeneratorBasedBuilder):
@@ -58,29 +56,24 @@ class Openwebtext(datasets.GeneratorBasedBuilder):
         )
 
     def _split_generators(self, dl_manager):
-        dl_dir = dl_manager.download_and_extract(_URL)
-        owt_dir = os.path.join(dl_dir, "openwebtext")
-        subset_xzs = [
-            os.path.join(owt_dir, file_name)
-            for file_name in sorted(os.listdir(owt_dir))
-            if file_name.endswith("xz")  # filter out ...xz.lock
-        ]
-        ex_dirs = dl_manager.extract(subset_xzs, num_proc=round(os.cpu_count() * 0.75))
-        nested_txt_files = [
-            [
-                os.path.join(ex_dir, txt_file_name)
-                for txt_file_name in sorted(os.listdir(ex_dir))
-                if txt_file_name.endswith("txt")
-            ]
-            for ex_dir in ex_dirs
-        ]
-        txt_files = chain(*nested_txt_files)
+        archives = dl_manager.download(_DATA_FILES)
         return [
-            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"txt_files": txt_files}),
+            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={
+                "archive_iterators": [
+                    dl_manager.iter_archive(archive) for archive in archives
+                ],
+                "iter_archive": dl_manager.iter_archive
+            }),
         ]
 
-    def _generate_examples(self, txt_files):
+    def _generate_examples(self, archive_iterators, iter_archive):
         """Yields examples."""
-        for idx, filepath in enumerate(txt_files):
-            with open(filepath, encoding="utf-8") as f:
-                yield idx, {"text": re.sub("\n\n\n+", "\n\n", f.read()).strip()}
+        for archive_iterator in archive_iterators:
+            for xz_filepath, xz_f in archive_iterator:
+                if not xz_filepath.endswith(".xz"):
+                    continue
+                for txt_filepath, txt_f in iter_archive(xz_f):
+                    if not txt_filepath.endswith(".txt"):
+                        continue
+                    idx = f"{xz_filepath}/{txt_filepath}"
+                    yield idx, {"text": re.sub("\n\n\n+", "\n\n", txt_f.read().decode("utf-8")).strip()}

subsets/urlsf_subset00.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cba8c76f8a27ea1d0d1430cfed59a84dc812e29d0e999d10305bc49848cd7127
+size 633047040

subsets/urlsf_subset01.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83891be18f14e9e2e7c7e7604c67f8ac44571f71bec34a6d57fe76e4bfe77c4c
+size 628838400

subsets/urlsf_subset02.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f6bfb3408b3aec9d22ef18ac84ee0fe34c8555366f9074a733dfa68d9ebf2d1
+size 629125120

subsets/urlsf_subset03.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69c91c02d754bfca9f8d9bddec43c5cac5cc728a62e09fd1a6ad66bef68ea900
+size 627578880

subsets/urlsf_subset04.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e9f206f3ab4d533a0f7a0d240b872d47a2b602d77ca09743fa02e5ee5ca31b1
+size 627189760

subsets/urlsf_subset05.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:49801cdaf398870eabd7225fce67d21afdd9923611c6fa16350ab39ed3ed5454
+size 630200320

subsets/urlsf_subset06.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78129d2aebbe5991a2b3b7568019aa080d56bef2bcd0571f3dde460f74ea8d7a
+size 625612800

subsets/urlsf_subset07.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d78f0c0d2890a20fa1f9f00d92959e1c450fe6b8c879e0596cccad9bc7aba17f
+size 625356800

subsets/urlsf_subset08.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17fecd74d654807b09d2b3328b8b71c050f2879bbe0f3514874be7f2fd3732ed
+size 624629760

subsets/urlsf_subset09.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65368771304487ba42d432d281d31e061679ceef5d05a4d0a9b3a0fae4add14f
+size 625807360

subsets/urlsf_subset10.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d06538c4ca1a1558c2d9305b86d853908c4ba5937e93e1ac74e57c32dfca8154
+size 625172480

subsets/urlsf_subset11.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:815dd5bfd712ff74465ad96114bbcd93a0e71e9b170aa06689e22d6f3b83f90d
+size 625264640

subsets/urlsf_subset12.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87619da8dfea1e843ff9644a08c52ce64bcb269383e1d22ac77483b65a5609cb
+size 624445440

subsets/urlsf_subset13.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db495f0255105e7515abab02bfe52d32b966159a86d3ce543656a0b768a59239
+size 628961280

subsets/urlsf_subset14.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae7dd7daa4fb6a4431115ded537569dbb521778d345a208f42184c9200f63352
+size 626708480

subsets/urlsf_subset15.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58c0aa0d3bd4139f9224b8f1b49577fe534d403afb825db92042fe6f5c5211b9
+size 620666880

subsets/urlsf_subset16.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0d3f20835741ce54d28577793c1f9adbd22d9c6593db55651d6f380c121e801
+size 618752000

subsets/urlsf_subset17.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0c20956c68bcba2a6cff7724db1a7c305c64a54f55ca5612f026beee51baa66
+size 619141120

subsets/urlsf_subset18.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e8970951bd8815ac18a1a29c0618847a4384b0803fe1e85e9a22606e63cf135d
+size 617789440

subsets/urlsf_subset19.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dfd91671633113ac5246334e217560ac57facb7e9353df5a56724c50a4e5c976
+size 619192320

subsets/urlsf_subset20.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1bc38f227abde149dbaee01e24eace67b534c1392b90655195f650b2b979ba52
+size 376709120