This dataset is provided by SAGE POND for research and development in Luganda language technology. To access it, Hugging Face will share your username and email address with SAGE POND. You may also optionally subscribe to updates about our products, services, research, and announcements.

By requesting access, you acknowledge that your Hugging Face username and email address will be shared with SAGE POND. Marketing and product communications are optional. You can unsubscribe from these communications at any time.

Luganda Grammar Error Correction Dataset

A synthetic dataset for training and evaluating Luganda Grammar Error Correction (GEC) models.

Dataset Description

Overview

The dataset consists of pairs of:

src: a corrupted Luganda sentence
tgt: the corresponding original/correct Luganda sentence

Corruptions are generated from clean Luganda text using linguistically informed corruption operations.

The objective is to train models to transform an erroneous Luganda sentence into its intended grammatical form.

Limitations

The dataset is synthetically corrupted and therefore does not represent the complete distribution of errors made by native or second-language Luganda writers.

Some corruption rules are deliberately conservative and only operate when the available linguistic information is sufficiently reliable.

The dataset should therefore be evaluated alongside naturally occurring learner-error data where available.

Citation

If you use this dataset in research or software, please cite SAGE POND.

@dataset{sagepond_luganda_gec,
  author       = {{SAGE POND}},
  title        = {Luganda Grammar Error Correction Dataset},
  year         = {2026},
  publisher    = {SAGE POND},
  license      = {ODC-By 1.0}
}

Downloads last month: 277

Total file size:

195 MB