Datasets:
mtanghu
commited on
Commit
•
382b6d2
1
Parent(s):
f3a6d99
Update Enwik8 broken link and information (#4950)
Browse files* Update enwik8 fixing the broken link
* Update enwik8 README file sizes
Commit from https://github.com/huggingface/datasets/commit/c7e540d2791a6eff7bdd5d5ecf029ed0da35802e
- README.md +9 -9
- dataset_infos.json +1 -1
- enwik8.py +3 -3
README.md
CHANGED
@@ -49,19 +49,19 @@ task_ids:
|
|
49 |
|
50 |
## Dataset Description
|
51 |
|
52 |
-
- **Homepage:**
|
53 |
- **Repository:** [Needs More Information]
|
54 |
- **Paper:** [Needs More Information]
|
55 |
-
- **Leaderboard:**
|
56 |
- **Point of Contact:** [Needs More Information]
|
57 |
|
58 |
### Dataset Summary
|
59 |
|
60 |
-
The enwik8
|
61 |
|
62 |
### Supported Tasks and Leaderboards
|
63 |
|
64 |
-
|
65 |
|
66 |
### Languages
|
67 |
|
@@ -71,9 +71,9 @@ en
|
|
71 |
|
72 |
### Data Instances
|
73 |
|
74 |
-
- **Size of downloaded dataset files:**
|
75 |
-
- **Size of generated dataset files:**
|
76 |
-
- **Total size:** 132.
|
77 |
|
78 |
```
|
79 |
{
|
@@ -110,7 +110,7 @@ The data fields are the same among all sets.
|
|
110 |
|
111 |
#### Initial Data Collection and Normalization
|
112 |
|
113 |
-
|
114 |
|
115 |
#### Who are the source language producers?
|
116 |
|
@@ -160,4 +160,4 @@ Dataset is not part of a publication, and can therefore not be cited.
|
|
160 |
|
161 |
### Contributions
|
162 |
|
163 |
-
Thanks to [@HallerPatrick](https://github.com/HallerPatrick) for adding this dataset.
|
|
|
49 |
|
50 |
## Dataset Description
|
51 |
|
52 |
+
- **Homepage:** http://mattmahoney.net/dc/textdata.html
|
53 |
- **Repository:** [Needs More Information]
|
54 |
- **Paper:** [Needs More Information]
|
55 |
+
- **Leaderboard:** https://paperswithcode.com/sota/language-modelling-on-enwiki8
|
56 |
- **Point of Contact:** [Needs More Information]
|
57 |
|
58 |
### Dataset Summary
|
59 |
|
60 |
+
The enwik8 dataset is the first 100,000,000 (100M) bytes of the English Wikipedia XML dump on Mar. 3, 2006 and is typically used to measure a model's ability to compress data.
|
61 |
|
62 |
### Supported Tasks and Leaderboards
|
63 |
|
64 |
+
A leaderboard for byte-level causal language modelling can be found on [paperswithcode](https://paperswithcode.com/sota/language-modelling-on-enwiki8)
|
65 |
|
66 |
### Languages
|
67 |
|
|
|
71 |
|
72 |
### Data Instances
|
73 |
|
74 |
+
- **Size of downloaded dataset files:** 34.76 MB
|
75 |
+
- **Size of generated dataset files:** 97.64 MB
|
76 |
+
- **Total size:** 132.40 MB
|
77 |
|
78 |
```
|
79 |
{
|
|
|
110 |
|
111 |
#### Initial Data Collection and Normalization
|
112 |
|
113 |
+
The data is just English Wikipedia XML dump on Mar. 3, 2006 split by line for enwik8 and not split by line for enwik8-raw.
|
114 |
|
115 |
#### Who are the source language producers?
|
116 |
|
|
|
160 |
|
161 |
### Contributions
|
162 |
|
163 |
+
Thanks to [@HallerPatrick](https://github.com/HallerPatrick) for adding this dataset and [@mtanghu](https://github.com/mtanghu) for updating it.
|
dataset_infos.json
CHANGED
@@ -1 +1 @@
|
|
1 |
-
{"enwik8": {"description": "The dataset is based on the Hutter Prize (http://prize.hutter1.net) and contains the first 10^8
|
|
|
1 |
+
{"enwik8": {"description": "The dataset is based on the Hutter Prize (http://prize.hutter1.net) and contains the first 10^8 bytes of English Wikipedia in 2006 in XML\n", "citation": "", "homepage": "http://mattmahoney.net/dc/textdata.html", "license": "", "features": {"text": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "enwik8", "config_name": "enwik8", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 104299244, "num_examples": 1128024, "dataset_name": "enwik8"}}, "download_checksums": {"http://mattmahoney.net/dc/enwik8.zip": {"num_bytes": 36445475, "checksum": "547994d9980ebed1288380d652999f38a14fe291a6247c157c3d33d4932534bc"}}, "download_size": 36445475, "post_processing_size": null, "dataset_size": 102383126, "size_in_bytes": 138828601}, "enwik8-raw": {"description": "The dataset is based on the Hutter Prize (http://prize.hutter1.net) and contains the first 10^8 bytes of English Wikipedia in 2006 in XML\n", "citation": "", "homepage": "http://mattmahoney.net/dc/textdata.html", "license": "", "features": {"text": {"dtype": "string", "id": null, "_type": "Value"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "enwik8", "config_name": "enwik8-raw", "version": {"version_str": "1.1.0", "description": null, "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 100000008, "num_examples": 1, "dataset_name": "enwik8"}}, "download_checksums": {"http://mattmahoney.net/dc/enwik8.zip": {"num_bytes": 36445475, "checksum": "547994d9980ebed1288380d652999f38a14fe291a6247c157c3d33d4932534bc"}}, "download_size": 36445475, "post_processing_size": null, "dataset_size": 100000008, "size_in_bytes": 136445483}}
|
enwik8.py
CHANGED
@@ -21,16 +21,16 @@ _CITATION = ""
|
|
21 |
|
22 |
# You can copy an official description
|
23 |
_DESCRIPTION = """\
|
24 |
-
The dataset is based on the Hutter Prize (http://prize.hutter1.net) and contains the first 10^8
|
25 |
"""
|
26 |
|
27 |
-
_HOMEPAGE = "
|
28 |
|
29 |
_LICENSE = ""
|
30 |
|
31 |
# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
|
32 |
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
|
33 |
-
_URLS = {"source": "http://
|
34 |
|
35 |
|
36 |
class Enwik8(datasets.GeneratorBasedBuilder):
|
|
|
21 |
|
22 |
# You can copy an official description
|
23 |
_DESCRIPTION = """\
|
24 |
+
The dataset is based on the Hutter Prize (http://prize.hutter1.net) and contains the first 10^8 bytes of English Wikipedia in 2006 in XML
|
25 |
"""
|
26 |
|
27 |
+
_HOMEPAGE = "http://mattmahoney.net/dc/textdata.html"
|
28 |
|
29 |
_LICENSE = ""
|
30 |
|
31 |
# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
|
32 |
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
|
33 |
+
_URLS = {"source": "http://mattmahoney.net/dc/enwik8.zip"}
|
34 |
|
35 |
|
36 |
class Enwik8(datasets.GeneratorBasedBuilder):
|