# Warrungu Chat Dataset

This dataset contains AI training prompts and responses designed to teach and reinforce the Warrungu Aboriginal language. It includes structured examples drawn from verified linguistic resources such as pronouns, suffix inventories, kinship terms, foods, animals, transitive/intransitive verbs, word order, and grammar rules.

The dataset supports instruction-tuned models like Mistral or Mixtral using frameworks such as Axolotl and Hugging Face Transformers.

## Structure

Each entry is formatted for supervised fine-tuning with instruction-following models:

```json
{
"prompt": "How do you say 'he is sitting' in Warrungu?",
"response": "Nyula nyina-n."
}