chuuhtetnaing/myanmar-instruction-tuning-dataset

Myanmar Instruction Tuning Dataset

A collection of question-answer pairs in Myanmar language for instruction tuning of Large Language Models.

Dataset Description

This dataset is a compilation of question-answer pairs in the Myanmar language, designed to support instruction tuning for Large Language Models. It aggregates content from multiple Myanmar language datasets covering diverse domains including agriculture, health, microbiology, general knowledge, and Buddhist teachings.

Dataset Structure

The dataset follows this structure:

• inputs: The question or instruction in Myanmar language
• targets: The corresponding answer or response
• source: The name of the original source dataset

Source Datasets

1. jojo-ai-mst/Myanmar-Agricutlure-1K: Agriculture Q&A dataset in Burmese language focusing on rice cultivation and agricultural practices in Myanmar.

2. jojo-ai-mst/Mpox-Myanmar: A dataset about Mpox (MonkeyPox virus) in Burmese language, containing fact-checked information from WHO and Myanmar government sources.

3. jojo-ai-mst/Burmese-Microbiology-1K: 1,000 question-answer pairs covering clinical microbiology in Burmese language.

4. chuuhtetnaing/myanmar-aya-dataset: A preprocessed subset of the Aya dataset containing only Myanmar language text.

5. chuuhtetnaing/dhamma-question-answer-dataset: Question-answer pairs derived from Dhamma articles, focusing on Buddhist teachings and principles.

6. thisisfalse9/Myanmar_Football_Dataset: A dataset containing football terminology, rules, and scenarios in Burmese language, designed for NLP research and LLM development focused on sports domain understanding.

Intended Use

This dataset aims to improve the performance of Large Language Models on Myanmar language understanding and instruction following. It can be used for:

• Fine-tuning LLMs to better support Myanmar language
• Creating language models that understand Burmese cultural and domain-specific contexts