Quentin Lhoest
commited on
Commit
•
05042a5
1
Parent(s):
ed847b4
Release: 2.0.0
Browse filesCommit from https://github.com/huggingface/datasets/commit/983f46ddae2f5b253db2b3c5691d38c75241cadb
README.md
CHANGED
@@ -670,15 +670,31 @@ The datasets are built from the Wikipedia dump
|
|
670 |
contains the content of one full Wikipedia article with cleaning to strip
|
671 |
markdown and unwanted sections (references, etc.).
|
672 |
|
673 |
-
The articles
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
674 |
|
675 |
### Supported Tasks and Leaderboards
|
676 |
|
677 |
-
|
678 |
|
679 |
### Languages
|
680 |
|
681 |
-
|
682 |
|
683 |
## Dataset Structure
|
684 |
|
@@ -686,6 +702,7 @@ We show detailed information for up to 5 configurations of the dataset.
|
|
686 |
|
687 |
### Data Instances
|
688 |
|
|
|
689 |
|
690 |
#### 20200501.en
|
691 |
|
@@ -809,10 +826,8 @@ Here are the sizes for several configurations:
|
|
809 |
title = "Wikimedia Downloads",
|
810 |
url = "https://dumps.wikimedia.org"
|
811 |
}
|
812 |
-
|
813 |
```
|
814 |
|
815 |
-
|
816 |
### Contributions
|
817 |
|
818 |
Thanks to [@lewtun](https://github.com/lewtun), [@mariamabarham](https://github.com/mariamabarham), [@thomwolf](https://github.com/thomwolf), [@lhoestq](https://github.com/lhoestq), [@patrickvonplaten](https://github.com/patrickvonplaten) for adding this dataset.
|
|
|
670 |
contains the content of one full Wikipedia article with cleaning to strip
|
671 |
markdown and unwanted sections (references, etc.).
|
672 |
|
673 |
+
The articles are parsed using the ``mwparserfromhell`` tool.
|
674 |
+
|
675 |
+
To load this dataset you need to install Apache Beam and ``mwparserfromhell`` first:
|
676 |
+
|
677 |
+
```
|
678 |
+
pip install apache_beam mwparserfromhell
|
679 |
+
```
|
680 |
+
|
681 |
+
Then can load any subset of Wikipedia per language and per date this way:
|
682 |
+
|
683 |
+
```python
|
684 |
+
from datasets import load_dataset
|
685 |
+
|
686 |
+
load_dataset("wikipedia", language="sw", date="20220120")
|
687 |
+
```
|
688 |
+
|
689 |
+
You can find the full list of languages and dates [here](https://dumps.wikimedia.org/backup-index.html).
|
690 |
|
691 |
### Supported Tasks and Leaderboards
|
692 |
|
693 |
+
The dataset is generally used for Language Modeling.
|
694 |
|
695 |
### Languages
|
696 |
|
697 |
+
You can find the list of languages [here](https://en.wikipedia.org/wiki/List_of_Wikipedias).
|
698 |
|
699 |
## Dataset Structure
|
700 |
|
|
|
702 |
|
703 |
### Data Instances
|
704 |
|
705 |
+
Some subsets of Wikipedia have already been processed by Hugging face, as you can see below:
|
706 |
|
707 |
#### 20200501.en
|
708 |
|
|
|
826 |
title = "Wikimedia Downloads",
|
827 |
url = "https://dumps.wikimedia.org"
|
828 |
}
|
|
|
829 |
```
|
830 |
|
|
|
831 |
### Contributions
|
832 |
|
833 |
Thanks to [@lewtun](https://github.com/lewtun), [@mariamabarham](https://github.com/mariamabarham), [@thomwolf](https://github.com/thomwolf), [@lhoestq](https://github.com/lhoestq), [@patrickvonplaten](https://github.com/patrickvonplaten) for adding this dataset.
|