Update README.md
Browse files
README.md
CHANGED
@@ -6,9 +6,9 @@ license: cc-by-nc-sa-4.0
|
|
6 |
|
7 |
## Dataset Description
|
8 |
|
9 |
-
Ko-LIMA는 Meta에서 공개한 [LIMA: Less Is More for Alignment (Zhou et al., 2023)](https://
|
10 |
|
11 |
-
데이터셋 관련하여 문의가 있으신
|
12 |
|
13 |
|
14 |
This is Korean LIMA dataset, which is translated from the [LIMA dataset](https://huggingface.co/datasets/GAIR/lima) that Meta's [LIMA model](https://arxiv.org/abs/2305.11206) (Zhou et al., 2023) was trained on. The translation has proceeded through [DeepL API](https://www.deepl.com/docs-api) with financial support from Tech Collaborative Lab in SK Inc.
|
|
|
6 |
|
7 |
## Dataset Description
|
8 |
|
9 |
+
Ko-LIMA는 Meta에서 공개한 [LIMA: Less Is More for Alignment](https://arxiv.org/abs/2305.11206) (Zhou et al., 2023)의 [학습 데이터](https://huggingface.co/datasets/GAIR/lima)를 한국어로 번역한 데이터셋입니다. 번역에는 [DeepL API](https://www.deepl.com/docs-api)를 활용하였고, SK(주) Tech Collaborative Lab으로부터 비용을 지원받았습니다. 전체 텍스트 중에서 code block이나 수식을 나타내는 특수문자 사이의 텍스트는 원문을 유지하는 형태로 번역을 진행하였으며, `train` 데이터셋 1,030건과 `test` 데이터셋 300건으로 구성된 총 1,330건의 데이터를 활용하실 수 있습니다. 현재 동일한 번역 문장을 `plain`, `vicuna` 두 가지 포멧으로 제공합니다.
|
10 |
|
11 |
+
데이터셋 관련하여 문의가 있으신 경우 [메일](mailto:[email protected])을 통해 연락주세요! 🥰
|
12 |
|
13 |
|
14 |
This is Korean LIMA dataset, which is translated from the [LIMA dataset](https://huggingface.co/datasets/GAIR/lima) that Meta's [LIMA model](https://arxiv.org/abs/2305.11206) (Zhou et al., 2023) was trained on. The translation has proceeded through [DeepL API](https://www.deepl.com/docs-api) with financial support from Tech Collaborative Lab in SK Inc.
|