Update README.md

README.md

@@ -2,22 +2,132 @@
 license: cc-by-sa-4.0
 task_categories:
 - text-generation
+- fill-mask
+task_ids:
+  - language-modeling
+  - masked-language-modeling
+configs:
+  - config_name: default
+    data_files:
+      - split: train
+        path: '20250320/*parquet/*parquet'
+  - config_name: ar
+    data_files:
+      - split: train
+        path: '20250320/ar*parquet/*parquet'
+  - config_name: br
+    data_files:
+      - split: train
+        path: '20250320/br*parquet/*parquet'
+  - config_name: ca
+    data_files:
+      - split: train
+        path: '20250320/ca*parquet/*parquet'
+  - config_name: co
+    data_files:
+      - split: train
+        path: '20250320/co*parquet/*parquet'
+  - config_name: de
+    data_files:
+      - split: train
+        path: '20250320/de*parquet/*parquet'
+  - config_name: en
+    data_files:
+      - split: train
+        path: '20250320/en*parquet/*parquet'
+  - config_name: es
+    data_files:
+      - split: train
+        path: '20250320/es*parquet/*parquet'
+  - config_name: eu
+    data_files:
+      - split: train
+        path: '20250320/eu*parquet/*parquet'
+  - config_name: fr
+    data_files:
+      - split: train
+        path: '20250320/fr*parquet/*parquet'
+  - config_name: it
+    data_files:
+      - split: train
+        path: '20250320/it*parquet/*parquet'
+  - config_name: nl
+    data_files:
+      - split: train
+        path: '20250320/nl*parquet/*parquet'
+  - config_name: oc
+    data_files:
+      - split: train
+        path: '20250320/oc*parquet/*parquet'
+  - config_name: pc
+    data_files:
+      - split: train
+        path: '20250320/pc*parquet/*parquet'
+  - config_name: pt
+    data_files:
+      - split: train
+        path: '20250320/pt*parquet/*parquet'
 language:
-- fr
 - ar
+- ba
+- br
+- ca
+- co
 - en
-- pt
-- nl
 - es
-- it
-- ba
 - eu
-- ca
+- fr
+- frp
+- it
+- nl
 - oc
-- co
-- br
+- pc
+- pt
 ---
 
+# Dataset Card
+
+This dataset is a curated collection of Wikimedia pages in markdown format,
+compiled from various Wikimedia projects across multiple languages.
+
+**Covered Wikimedia Projects:**
+* wikipedia
+* wikibooks
+* wikinews
+* wikiquote
+* wikisource
+* wikiversity
+* wikivoyage
+* wiktionary
+
+**Supported Languages:**
+* ar (Arabic)
+* br (Breton)
+* ca (Catalan)
+* co (Corsican)
+* de (German)
+* en (English)
+* es (Spanish)
+* eu (Basque)
+* fr (French)
+* frp (Arpitan)
+* it (Italian)
+* nl (Dutch)
+* oc (Occitan)
+* pcd (Picard)
+* pt (Portuguese)
+
+## Data Source
+
+The content was extracted from the  [Wikimedia dumps](https://dumps.wikimedia.org/other/enterprise_html/runs),
+specifically from the dump dated March 20, 2025 (20250320).
+
+The extraction process follows the same methodology as the one used in the [OpenLLM-France/wikipedia](https://huggingface.co/datasets/OpenLLM-France/wikipedia).
+
+## Dataset Size
+
+The tables below provide a detailed overview of the dataset size, organized by language and Wikimedia project.
+
 | language | domain | # pages | # words | # characters |
 |----------|--------|----------|-----------|-----------|
 | ar | wikipedia | 1.29 M | 448.80 M | 2.60 B |
@@ -68,7 +178,7 @@ language:
 | eu | wikiquote | 370 | 90.20 K | 654.51 K |
 | eu | wikisource | 1.24 K | 3.61 M | 26.83 M |
 | eu | wiktionary | 54.94 K | 3.08 M | 23.94 M |
-| fr | pwiki | 5.79 K | 873.34 K | 4.97 M |
+| frp | wikipedia | 5.79 K | 873.34 K | 4.97 M |
 | fr | wikipedia | 2.77 M | 1.89 B | 10.90 B |
 | fr | wikibooks | 23.74 K | 26.79 M | 161.96 M |
 | fr | wikinews | 24.26 K | 12.03 M | 63.53 M |
@@ -95,7 +205,7 @@ language:
 | oc | wikipedia | 87.87 K | 30.29 M | 165.44 M |
 | oc | wikibooks | 67 | 22.36 K | 127.57 K |
 | oc | wiktionary | 72.21 K | 5.63 M | 37.30 M |
-| pc | dwiki | 6.04 K | 1.59 M | 8.92 M |
+| pcd | wikipedia | 6.04 K | 1.59 M | 8.92 M |
 | pt | wikipedia | 1.17 M | 617.13 M | 3.52 B |
 | pt | wikibooks | 11.74 K | 7.87 M | 49.26 M |
 | pt | wikinews | 34.54 K | 16.16 M | 91.11 M |
@@ -113,9 +223,31 @@ language:
 | es (Spanish) | total | 3.06 M | 1.56 B | 9.07 B |
 | eu (Basque) | total | 511.70 K | 124.81 M | 882.24 M |
 | fr (French) | total | 9.66 M | 3.07 B | 18.00 B |
+| frp (Arpitan) | total | 5.79 K | 873.34 K | 4.97 M |
 | it (Italian) | total | 2.75 M | 1.48 B | 8.86 B |
 | nl (Dutch) | total | 3.16 M | 734.36 M | 4.40 B |
 | oc (Occitan) | total | 160.15 K | 35.94 M | 202.87 M |
-| pc (Picard) | total | 6.04 K | 1.59 M | 8.92 M |
+| pcd (Picard) | total | 6.04 K | 1.59 M | 8.92 M |
 | pt (Portuguese) | total | 1.76 M | 710.99 M | 4.06 B |
 | Total | | 45.16 M | 18.03 B | 107.07 B |
+
+## Example use (python)
+
+Load the full dataset:
+
+```python
+import datasets
+
+ds = datasets.load_dataset("OpenLLM-France/wikimedia",
+    streaming=True,
+    split="train"
+)
+```
+
+Load the dataset for a given language (French in this example):
+```python
+ds = datasets.load_dataset("OpenLLM-France/wikimedia", "fr",
+    streaming=True,
+    split="train"
+)
+```