Datasets:
Tasks:
Text Generation
Modalities:
Text
Formats:
json
Languages:
Indonesian
Size:
1K - 10K
Tags:
instruction-finetuning
metadata
language:
- id
tags:
- instruction-finetuning
pretty_name: Betawi v0
task_categories:
- text-generation
Synthetic Betawi Language dataset, generated by GPT-4o: Betawi v0 (Alpha)
Betawi v0 is a synthetic dataset created using GPT-4o, consisting of over 1,000 instruction-output pairs across a range of topics, structured in JSON format. It follows the Alpaca dataset format and is designed for fine-tuning large language models (LLMs) to enhance LLMs understanding of Bahasa Betawi.
Version Alpha