---
dataset_info:
  features:
  - name: text
    dtype: string
  - name: source_dataset
    dtype: string
  splits:
  - name: train
    num_bytes: 21395236.84555232
    num_examples: 84328
  download_size: 43499142
  dataset_size: 21395236.84555232
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
task_categories:
- text-generation
- mask-generation
language:
- bm
size_categories:
- 10K<n<100K
---

## Dataset Summary

The **Bambara-Texts** dataset is a collection of monolingual Bambara text designed for **pretraining language models**. It provides a diverse set of textual data to improve natural language processing (NLP) applications for the **Bambara language**. 

This dataset can be used for:
- **Pretraining large language models (LLMs)**
- **Building word embeddings for Bambara**
- **Language modeling tasks such as masked language modeling (MLM) and autoregressive modeling**
- **Corpus-based linguistic research on Bambara**

## Project

This dataset is part of a broader initiative to advance **Bambara NLP** and bridge the gap in AI accessibility for low-resource languages. By providing high-quality monolingual data, we aim to **enable better machine translation, speech recognition, and text generation models for Bambara**.

## Bambara Language

Bambara, also known as **Bamanankan**, is a Mande language spoken primarily in **Mali** by millions of people as both a first and second language. It serves as a **lingua franca** in Mali and is also spoken in parts of **Burkina Faso, Ivory Coast, and Senegal**. 

Bambara is traditionally written using the **Latin script**, though **N'Ko script** is also widely used. It has a strong oral tradition and plays a vital role in Malian cultural heritage.


### Structure
The dataset consists of:
- `text`: Raw text samples in Bambara.
- `source_dataset`: The original dataset or project from which the text was sourced.


## Citation
This dataset has been curated by [oza75](https://huggingface.co/oza75) at Djelia.

If you use this dataset in your research or application, please cite it as:

```
@misc{bambara-texts,
  author = {Djelia},
  title = {Bambara-Texts Dataset},
  year = {2025},
  url = {https://huggingface.co/datasets/djelia/bambara-texts}
}
```