---
dataset_info:
  features:
  - name: sentence
    dtype: string
  - name: length
    dtype: int64
  splits:
  - name: train
    num_bytes: 830143889
    num_examples: 3700048
  - name: test
    num_bytes: 207492305
    num_examples: 925012
  download_size: 499982684
  dataset_size: 1037636194
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
license: apache-2.0
language:
- dv
tags:
- thaana
- dhivehi
- sentences
- normalized
- spoken
pretty_name: thaana_sentences
size_categories:
- 1M<n<10M
---

# Dhivehi Sentences Dataset

This repository contains a dataset of Dhivehi (Thaana, Maldivian) sentences processed and normalized for natural language processing tasks.

***This is an extended version of the [Dhivehi Sentences Dataset](https://huggingface.co/datasets/alakxender/dhivehi-sentences-normalized)***

## Dataset Description

The dataset combines text from multiple sources:

- Random News Articles
- Glot500 Dhivehi-Thaana
- FineWeb-2 Dhivehi-Thaana

The sentences have been processed to:

- Split into individual sentences
- Normalize numbers into Dhivehi text form
- Normalize years into proper Dhivehi format
- Normalize time expressions
- Convert sentence endings to spoken form
- Standardize common suffixes and abbreviations

## Dataset Format

The dataset is split into train (80%) and test (20%) sets. Each row contains:

- sentence: The processed Dhivehi text
- length: Character length of the sentence

## Usage

```python
from datasets import load_dataset

ds = load_dataset("alakxender/dhivehi-sentences-extended")

# Print the first 10 rows of the dataset
print(ds["train"][:10])
```

---

***Note: Chunked and created using [dv-normalizer](https://pypi.org/project/dv-normalizer/) v0.1.2***