---
dataset_info:
  features:
  - name: text
    dtype: string
  splits:
  - name: train
    num_bytes: 6242898204.747201
    num_examples: 2539940
  - name: validation
    num_bytes: 328575918.2527994
    num_examples: 133682
  download_size: 3992353312
  dataset_size: 6571474123
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
language:
- fr
---
# Dataset Card for French Wikipedia Text Corpus
## Dataset Description
The French Wikipedia Text Corpus is a comprehensive dataset derived from French Wikipedia articles. It is specifically designed for training language models (LLMs). The dataset contains the text of paragraphs from Wikipedia articles, with sections, footnotes, and titles removed to provide a clean and continuous text stream.
Dataset Details
Features

    text: A single attribute containing the full text of Wikipedia articles as a continuous chunk of text.

## Languages
The dataset is in French.

## Dataset statistics:
Train set: 2539940 examples
Validation set: 133682 examples

## Dataset Creation
The dataset was created by extracting and processing text from French Wikipedia articles. The processing involved removing article sections, footnotes, and titles to ensure that the text is in a format suitable for training language models.

## Intended Uses
This dataset is intended for use in training and evaluating language models. The clean and continuous text format makes it particularly suitable for models that require large volumes of coherent text data.

## Example Usage
```python
from datasets import load_dataset

# Load the dataset
dataset = load_dataset('1ou2/fr_wiki_paragraphs')

# Access the training data
train_data = dataset['train']

# Print an example
print(train_data[0]['text'])
```

## Citation Information
If you use this dataset in your research, please cite it as follows:

@misc{french_wikipedia_text_corpus,
  author = {Gabriel Pastor},
  title = {French Wikipedia Text Corpus},
  year = {2025},
  publisher = {Hugging Face},
  journal = {Hugging Face Datasets},
  howpublished = {\url{https://huggingface.co/datasets/1ou2/fr_wiki_paragraphs}}
}

## License
This dataset is licensed under the CC BY-SA 3.0 license, in accordance with Wikipedia's content license.