README.md

---
license: mit
datasets:
- nlpai-lab/kullm-v2
- maywell/ko_wikidata_QA
- eaglewatch/Korean_Wikipedia_Dataset_for_GPT2_August_2022
language:
- ko
---

[kogpt-j-base](https://huggingface.co/heegyu/kogpt-j-base)모델을 데이터셋 [세](https://huggingface.co/datasets/nlpai-lab/kullm-v2)[가](https://huggingface.co/datasets/maywell/ko_wikidata_QA)[지](https://huggingface.co/datasets/eaglewatch/Korean_Wikipedia_Dataset_for_GPT2_August_2022)를 사용해서 45k step(0.27 epoch)만큼 학습시킨 모델입니다.


프롬프트:

```
<|im_start|>system
당신은 사람들을 도와주는 인공지능 비서입니다. 질문을 읽고 알맞은 답변을 제공하세요.<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
```


데이터셋:

[nlpai-lab/kullm-v2](https://huggingface.co/datasets/nlpai-lab/kullm-v2)

[maywell/ko_wikidata_QA](https://huggingface.co/datasets/maywell/ko_wikidata_QA)

[eaglewatch/Korean_Wikipedia_Dataset_for_GPT2_August_2022](https://huggingface.co/datasets/eaglewatch/Korean_Wikipedia_Dataset_for_GPT2_August_2022)