---
dataset_info:
  features:
  - name: incorrect
    dtype: string
  - name: correct
    dtype: string
  splits:
  - name: train
    num_bytes: 193960468
    num_examples: 540000
  - name: test
    num_bytes: 21546211
    num_examples: 60000
  download_size: 129134550
  dataset_size: 215506679
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
language:
- ur
---


This dataset contains paired Urdu text for error correction tasks. 

## Data Format
Each dialogue entry in the dataset is structured as follows:

- `input_text`:  The text containing simulated errors.  
- `correct_text`:  The corresponding correct text.