---
license: mit
language:
  - km
  - en
pretty_name: Khmer Food Dataset
tags:
  - khmer
  - cambodian-food
  - nlp
  - instruction-dataset
---

# Khmer Food Dataset

Khmer Food Dataset is a small educational dataset containing structured
information and question-answer examples about Cambodian food.

This dataset was created for learning purposes while building a small
Khmer language model from scratch.

## Dataset

The current dataset contains 40 Cambodian food entries.

Each food may contain:

- Khmer name
- English name
- category
- description
- ingredients
- taste
- meal type

## Splits

| Split | Foods | Examples |
|---|---:|---:|
| Train | 28 | 418 |
| Validation | 6 | 86 |
| Test | 6 | 90 |

The split is performed at the food level before question augmentation to
reduce data leakage between training and evaluation.

## Purpose

This dataset is intended for learning and educational experiments with:

- Khmer NLP
- Tokenization
- Language modeling
- Instruction datasets
- Transformer training

## Author

Chhorn Seyha