Skip to content

Dataset · squarelike

Korean translation of OpenOrca instructions

squarelike/OpenOrca-gugugo-ko

Over 2.2 million instruction examples in Korean for SFT model training.

What's inside

A Korean text dataset representing a translation of the well-known OpenOrca corpus, performed using the Gugugo-koen-7B-V1.1 model. It includes about 640,000 translated GPT-4 answers and about 1.59 million GPT-3.5 answers (total size 1 to 10 million entries). Suitable for a wide range of NLP tasks, including question answering, summarization, and classification.

What you can do with it

  • SFT fine-tuning of language models in Korean
  • Training models for summarization and table-based QA
  • Training text classifiers on Korean data
Tasks
Text classification, Token classification, Table question answering, Question answering, Zero shot classification, Summarization
Languages
Korean
Size
1M–10M rows
Downloads this month
137
Likes
35

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("squarelike/OpenOrca-gugugo-ko")