Quentin Lhoest
commited on
Commit
·
858f724
1
Parent(s):
04eb92e
Release: 2.0.0
Browse filesCommit from https://github.com/huggingface/datasets/commit/983f46ddae2f5b253db2b3c5691d38c75241cadb
README.md
CHANGED
@@ -686,15 +686,31 @@ The datasets are built from the Wikipedia dump
|
|
686 |
contains the content of one full Wikipedia article with cleaning to strip
|
687 |
markdown and unwanted sections (references, etc.).
|
688 |
|
689 |
-
The articles
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
690 |
|
691 |
### Supported Tasks and Leaderboards
|
692 |
|
693 |
-
|
694 |
|
695 |
### Languages
|
696 |
|
697 |
-
|
698 |
|
699 |
## Dataset Structure
|
700 |
|
@@ -702,6 +718,7 @@ We show detailed information for up to 5 configurations of the dataset.
|
|
702 |
|
703 |
### Data Instances
|
704 |
|
|
|
705 |
|
706 |
#### 20200501.en
|
707 |
|
@@ -825,10 +842,8 @@ Here are the sizes for several configurations:
|
|
825 |
title = "Wikimedia Downloads",
|
826 |
url = "https://dumps.wikimedia.org"
|
827 |
}
|
828 |
-
|
829 |
```
|
830 |
|
831 |
-
|
832 |
### Contributions
|
833 |
|
834 |
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.
|
|
|
686 |
contains the content of one full Wikipedia article with cleaning to strip
|
687 |
markdown and unwanted sections (references, etc.).
|
688 |
|
689 |
+
The articles are parsed using the ``mwparserfromhell`` tool.
|
690 |
+
|
691 |
+
To load this dataset you need to install Apache Beam and ``mwparserfromhell`` first:
|
692 |
+
|
693 |
+
```
|
694 |
+
pip install apache_beam mwparserfromhell
|
695 |
+
```
|
696 |
+
|
697 |
+
Then can load any subset of Wikipedia per language and per date this way:
|
698 |
+
|
699 |
+
```python
|
700 |
+
from datasets import load_dataset
|
701 |
+
|
702 |
+
load_dataset("wikipedia", language="sw", date="20220120")
|
703 |
+
```
|
704 |
+
|
705 |
+
You can find the full list of languages and dates [here](https://dumps.wikimedia.org/backup-index.html).
|
706 |
|
707 |
### Supported Tasks and Leaderboards
|
708 |
|
709 |
+
The dataset is generally used for Language Modeling.
|
710 |
|
711 |
### Languages
|
712 |
|
713 |
+
You can find the list of languages [here](https://en.wikipedia.org/wiki/List_of_Wikipedias).
|
714 |
|
715 |
## Dataset Structure
|
716 |
|
|
|
718 |
|
719 |
### Data Instances
|
720 |
|
721 |
+
Some subsets of Wikipedia have already been processed by Hugging face, as you can see below:
|
722 |
|
723 |
#### 20200501.en
|
724 |
|
|
|
842 |
title = "Wikimedia Downloads",
|
843 |
url = "https://dumps.wikimedia.org"
|
844 |
}
|
|
|
845 |
```
|
846 |
|
|
|
847 |
### Contributions
|
848 |
|
849 |
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.
|