Update files from the datasets library (from 1.16.0)
Browse filesRelease notes: https://github.com/huggingface/datasets/releases/tag/1.16.0
- README.md +33 -7
- dataset_infos.json +1 -1
- wikitext.py +25 -21
README.md
CHANGED
@@ -1,7 +1,25 @@
|
|
1 |
---
|
|
|
|
|
|
|
|
|
2 |
languages:
|
3 |
- en
|
|
|
|
|
|
|
|
|
|
|
4 |
paperswithcode_id: wikitext-2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
---
|
6 |
|
7 |
# Dataset Card for "wikitext"
|
@@ -34,8 +52,8 @@ paperswithcode_id: wikitext-2
|
|
34 |
|
35 |
- **Homepage:** [https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/](https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/)
|
36 |
- **Repository:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
37 |
-
- **Paper:** [
|
38 |
-
- **Point of Contact:** [
|
39 |
- **Size of downloaded dataset files:** 373.28 MB
|
40 |
- **Size of the generated dataset:** 1072.25 MB
|
41 |
- **Total amount of disk used:** 1445.53 MB
|
@@ -45,6 +63,11 @@ paperswithcode_id: wikitext-2
|
|
45 |
The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified
|
46 |
Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike License.
|
47 |
|
|
|
|
|
|
|
|
|
|
|
48 |
### Supported Tasks and Leaderboards
|
49 |
|
50 |
[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
@@ -196,16 +219,19 @@ The data fields are the same among all splits.
|
|
196 |
|
197 |
### Licensing Information
|
198 |
|
199 |
-
[
|
200 |
|
201 |
### Citation Information
|
202 |
|
203 |
```
|
204 |
-
@
|
205 |
-
|
206 |
-
|
|
|
|
|
|
|
|
|
207 |
}
|
208 |
-
|
209 |
```
|
210 |
|
211 |
|
|
|
1 |
---
|
2 |
+
annotations_creators:
|
3 |
+
- no-annotation
|
4 |
+
language_creators:
|
5 |
+
- crowdsourced
|
6 |
languages:
|
7 |
- en
|
8 |
+
licenses:
|
9 |
+
- cc-by-sa-3-0
|
10 |
+
- gfdl-1-3-or-later
|
11 |
+
multilinguality:
|
12 |
+
- monolingual
|
13 |
paperswithcode_id: wikitext-2
|
14 |
+
pretty_name: WikiText
|
15 |
+
size_categories:
|
16 |
+
- 1M<n<10M
|
17 |
+
source_datasets:
|
18 |
+
- original
|
19 |
+
task_categories:
|
20 |
+
- sequence-modeling
|
21 |
+
task_ids:
|
22 |
+
- language-modeling
|
23 |
---
|
24 |
|
25 |
# Dataset Card for "wikitext"
|
|
|
52 |
|
53 |
- **Homepage:** [https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/](https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/)
|
54 |
- **Repository:** [More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
55 |
+
- **Paper:** [Pointer Sentinel Mixture Models](https://arxiv.org/abs/1609.07843)
|
56 |
+
- **Point of Contact:** [Stephen Merity](mailto:smerity@salesforce.com)
|
57 |
- **Size of downloaded dataset files:** 373.28 MB
|
58 |
- **Size of the generated dataset:** 1072.25 MB
|
59 |
- **Total amount of disk used:** 1445.53 MB
|
|
|
63 |
The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified
|
64 |
Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike License.
|
65 |
|
66 |
+
Compared to the preprocessed version of Penn Treebank (PTB), WikiText-2 is over 2 times larger and WikiText-103 is over
|
67 |
+
110 times larger. The WikiText dataset also features a far larger vocabulary and retains the original case, punctuation
|
68 |
+
and numbers - all of which are removed in PTB. As it is composed of full articles, the dataset is well suited for models
|
69 |
+
that can take advantage of long term dependencies.
|
70 |
+
|
71 |
### Supported Tasks and Leaderboards
|
72 |
|
73 |
[More Information Needed](https://github.com/huggingface/datasets/blob/master/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
|
|
|
219 |
|
220 |
### Licensing Information
|
221 |
|
222 |
+
The dataset is available under the [Creative Commons Attribution-ShareAlike License (CC BY-SA 4.0)](https://creativecommons.org/licenses/by-sa/4.0/).
|
223 |
|
224 |
### Citation Information
|
225 |
|
226 |
```
|
227 |
+
@misc{merity2016pointer,
|
228 |
+
title={Pointer Sentinel Mixture Models},
|
229 |
+
author={Stephen Merity and Caiming Xiong and James Bradbury and Richard Socher},
|
230 |
+
year={2016},
|
231 |
+
eprint={1609.07843},
|
232 |
+
archivePrefix={arXiv},
|
233 |
+
primaryClass={cs.CL}
|
234 |
}
|
|
|
235 |
```
|
236 |
|
237 |
|
dataset_infos.json
CHANGED
@@ -1 +1 @@
|
|
1 |
-
{"wikitext-103-
|
|
|
1 |
+
{"wikitext-103-v1": {"description": " The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified\n Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike\n License.\n", "citation": "@misc{merity2016pointer,\n title={Pointer Sentinel Mixture Models},\n author={Stephen Merity and Caiming Xiong and James Bradbury and Richard Socher},\n year={2016},\n eprint={1609.07843},\n archivePrefix={arXiv},\n primaryClass={cs.CL}\n}\n", "homepage": "https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/", "license": "Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)", "features": {"text": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wikitext", "config_name": "wikitext-103-v1", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1295579, "num_examples": 4358, "dataset_name": "wikitext"}, "train": {"name": "train", "num_bytes": 545142639, "num_examples": 1801350, "dataset_name": "wikitext"}, "validation": {"name": "validation", "num_bytes": 1154755, "num_examples": 3760, "dataset_name": "wikitext"}}, "download_checksums": {"https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-103-v1.zip": {"num_bytes": 190229076, "checksum": "242ba0f20b329cfdf1ccc61e9e9e5b59becf189db7f7a81cd2a0e2fc31539590"}}, "download_size": 190229076, "post_processing_size": null, "dataset_size": 547592973, "size_in_bytes": 737822049}, "wikitext-2-v1": {"description": " The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified\n Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike\n License.\n", "citation": "@misc{merity2016pointer,\n title={Pointer Sentinel Mixture Models},\n author={Stephen Merity and Caiming Xiong and James Bradbury and Richard Socher},\n year={2016},\n eprint={1609.07843},\n archivePrefix={arXiv},\n primaryClass={cs.CL}\n}\n", "homepage": "https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/", "license": "Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)", "features": {"text": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wikitext", "config_name": "wikitext-2-v1", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1270951, "num_examples": 4358, "dataset_name": "wikitext"}, "train": {"name": "train", "num_bytes": 10918134, "num_examples": 36718, "dataset_name": "wikitext"}, "validation": {"name": "validation", "num_bytes": 1134127, "num_examples": 3760, "dataset_name": "wikitext"}}, "download_checksums": {"https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-2-v1.zip": {"num_bytes": 4475746, "checksum": "92675f1d63015c1c8b51f1656a52d5bdbc33aafa60cc47a218a66e7ee817488c"}}, "download_size": 4475746, "post_processing_size": null, "dataset_size": 13323212, "size_in_bytes": 17798958}, "wikitext-103-raw-v1": {"description": " The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified\n Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike\n License.\n", "citation": "@misc{merity2016pointer,\n title={Pointer Sentinel Mixture Models},\n author={Stephen Merity and Caiming Xiong and James Bradbury and Richard Socher},\n year={2016},\n eprint={1609.07843},\n archivePrefix={arXiv},\n primaryClass={cs.CL}\n}\n", "homepage": "https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/", "license": "Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)", "features": {"text": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wikitext", "config_name": "wikitext-103-raw-v1", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1305092, "num_examples": 4358, "dataset_name": "wikitext"}, "train": {"name": "train", "num_bytes": 546501673, "num_examples": 1801350, "dataset_name": "wikitext"}, "validation": {"name": "validation", "num_bytes": 1159292, "num_examples": 3760, "dataset_name": "wikitext"}}, "download_checksums": {"https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-103-raw-v1.zip": {"num_bytes": 191984949, "checksum": "91c00ae287f0d699e18605c84afc9e45c192bc6b7797ff8837e5474655a33794"}}, "download_size": 191984949, "post_processing_size": null, "dataset_size": 548966057, "size_in_bytes": 740951006}, "wikitext-2-raw-v1": {"description": " The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified\n Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike\n License.\n", "citation": "@misc{merity2016pointer,\n title={Pointer Sentinel Mixture Models},\n author={Stephen Merity and Caiming Xiong and James Bradbury and Richard Socher},\n year={2016},\n eprint={1609.07843},\n archivePrefix={arXiv},\n primaryClass={cs.CL}\n}\n", "homepage": "https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/", "license": "Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)", "features": {"text": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wikitext", "config_name": "wikitext-2-raw-v1", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 1305092, "num_examples": 4358, "dataset_name": "wikitext"}, "train": {"name": "train", "num_bytes": 11061733, "num_examples": 36718, "dataset_name": "wikitext"}, "validation": {"name": "validation", "num_bytes": 1159292, "num_examples": 3760, "dataset_name": "wikitext"}}, "download_checksums": {"https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-2-raw-v1.zip": {"num_bytes": 4721645, "checksum": "ef7edb566e3e2b2d31b29c1fdb0c89a4cc683597484c3dc2517919c615435a11"}}, "download_size": 4721645, "post_processing_size": null, "dataset_size": 13526117, "size_in_bytes": 18247762}}
|
wikitext.py
CHANGED
@@ -6,20 +6,24 @@ import os
|
|
6 |
import datasets
|
7 |
|
8 |
|
9 |
-
# TODO(wikitext): BibTeX citation
|
10 |
_CITATION = """\
|
11 |
-
@
|
12 |
-
|
13 |
-
|
|
|
|
|
|
|
|
|
14 |
}
|
15 |
"""
|
16 |
|
17 |
-
# TODO(wikitext):
|
18 |
_DESCRIPTION = """\
|
19 |
The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified
|
20 |
-
Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike
|
|
|
21 |
"""
|
22 |
-
|
|
|
23 |
_DATA_URL = "https://s3.amazonaws.com/research.metamind.io/wikitext"
|
24 |
|
25 |
|
@@ -49,25 +53,25 @@ class Wikitext(datasets.GeneratorBasedBuilder):
|
|
49 |
VERSION = datasets.Version("0.1.0")
|
50 |
BUILDER_CONFIGS = [
|
51 |
WikitextConfig(
|
52 |
-
name="wikitext-103-
|
53 |
-
data_url=_DATA_URL + "/" + "wikitext-103-
|
54 |
-
description="
|
55 |
),
|
56 |
WikitextConfig(
|
57 |
-
name="wikitext-2-
|
58 |
-
data_url=_DATA_URL + "/" + "wikitext-2-
|
59 |
-
description="
|
60 |
),
|
61 |
WikitextConfig(
|
62 |
-
name="wikitext-103-v1",
|
63 |
-
data_url=_DATA_URL + "/" + "wikitext-103-v1.zip",
|
64 |
-
description="
|
65 |
"They should only be used for character level work or for creating newly derived datasets.",
|
66 |
),
|
67 |
WikitextConfig(
|
68 |
-
name="wikitext-2-v1",
|
69 |
-
data_url=_DATA_URL + "/" + "wikitext-2-v1.zip",
|
70 |
-
description="
|
71 |
"They should only be used for character level work or for creating newly derived datasets.",
|
72 |
),
|
73 |
]
|
@@ -88,8 +92,8 @@ class Wikitext(datasets.GeneratorBasedBuilder):
|
|
88 |
# specify them here. They'll be used if as_supervised=True in
|
89 |
# builder.as_dataset.
|
90 |
supervised_keys=None,
|
91 |
-
|
92 |
-
|
93 |
citation=_CITATION,
|
94 |
)
|
95 |
|
|
|
6 |
import datasets
|
7 |
|
8 |
|
|
|
9 |
_CITATION = """\
|
10 |
+
@misc{merity2016pointer,
|
11 |
+
title={Pointer Sentinel Mixture Models},
|
12 |
+
author={Stephen Merity and Caiming Xiong and James Bradbury and Richard Socher},
|
13 |
+
year={2016},
|
14 |
+
eprint={1609.07843},
|
15 |
+
archivePrefix={arXiv},
|
16 |
+
primaryClass={cs.CL}
|
17 |
}
|
18 |
"""
|
19 |
|
|
|
20 |
_DESCRIPTION = """\
|
21 |
The WikiText language modeling dataset is a collection of over 100 million tokens extracted from the set of verified
|
22 |
+
Good and Featured articles on Wikipedia. The dataset is available under the Creative Commons Attribution-ShareAlike
|
23 |
+
License.
|
24 |
"""
|
25 |
+
_HOMEPAGE = "https://blog.einstein.ai/the-wikitext-long-term-dependency-language-modeling-dataset/"
|
26 |
+
_LICENSE = "Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)"
|
27 |
_DATA_URL = "https://s3.amazonaws.com/research.metamind.io/wikitext"
|
28 |
|
29 |
|
|
|
53 |
VERSION = datasets.Version("0.1.0")
|
54 |
BUILDER_CONFIGS = [
|
55 |
WikitextConfig(
|
56 |
+
name="wikitext-103-v1",
|
57 |
+
data_url=_DATA_URL + "/" + "wikitext-103-v1.zip",
|
58 |
+
description="Word level dataset. No processing is needed other than replacing newlines with <eos> tokens.",
|
59 |
),
|
60 |
WikitextConfig(
|
61 |
+
name="wikitext-2-v1",
|
62 |
+
data_url=_DATA_URL + "/" + "wikitext-2-v1.zip",
|
63 |
+
description="Word level dataset. No processing is needed other than replacing newlines with <eos> tokens.",
|
64 |
),
|
65 |
WikitextConfig(
|
66 |
+
name="wikitext-103-raw-v1",
|
67 |
+
data_url=_DATA_URL + "/" + "wikitext-103-raw-v1.zip",
|
68 |
+
description="Raw level dataset: the raw tokens before the addition of <unk> tokens. "
|
69 |
"They should only be used for character level work or for creating newly derived datasets.",
|
70 |
),
|
71 |
WikitextConfig(
|
72 |
+
name="wikitext-2-raw-v1",
|
73 |
+
data_url=_DATA_URL + "/" + "wikitext-2-raw-v1.zip",
|
74 |
+
description="Raw level dataset: the raw tokens before the addition of <unk> tokens. "
|
75 |
"They should only be used for character level work or for creating newly derived datasets.",
|
76 |
),
|
77 |
]
|
|
|
92 |
# specify them here. They'll be used if as_supervised=True in
|
93 |
# builder.as_dataset.
|
94 |
supervised_keys=None,
|
95 |
+
homepage=_HOMEPAGE,
|
96 |
+
license=_LICENSE,
|
97 |
citation=_CITATION,
|
98 |
)
|
99 |
|