Datasets:
Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html>
<h"... is not valid JSON
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Pashto Corpus Training Ready
Cleaned and chunked Pashto text prepared for language model training and evaluation. The repository stores Arrow shards for train, validation, and test.
Fields
text: cleaned Pashto text chunksource: source categorylength: character count
Splits
| Split | Purpose |
|---|---|
train |
model training |
validation |
tuning and model selection |
test |
final evaluation |
Usage
from datasets import load_dataset
dataset = load_dataset("tasal9/Pashto-Corpus-Training-Ready")
print(dataset)
Processing
- Removed PDF artifacts and page numbers
- Normalized Arabic/Pashto characters
- Split long documents into chunks
- Filtered short or noisy samples
- Created train/validation/test splits
License
CC BY 4.0.
- Downloads last month
- 79