concept_set
stringclasses 5
values | 1
stringclasses 5
values | 2
stringclasses 5
values | 3
stringclasses 5
values | 4
stringclasses 5
values | gold
int64 1
4
|
---|---|---|---|---|---|
나#교훈적#내용#주제#강연#하다 | 나는 교훈적인 내용이 강연을 하다. | 나는 교훈적인 내용을 주제로 강연을 하지 않았다. | 나는 교훈적인 내용이 주제 때문에 강연을 했어. | 교훈적인 내용이 나를 강연에게 하다. | 2 |
옳다#그르다#판단하다 | 나는 옳고 그름을 객관적으로 판단하지 않았다. | 나는 옳고 그름을 객관적으로 안 해야 했다. | 옳고 그름을 판단하게 객관적이었다. | 옳고 그름이 누가 판단하게 정당하게 봤다. | 1 |
시민#단체#환경#정책#반대하다#파괴하다 | 시민 단체는 환경을 파괴하는 정책을 반대하는 척 한다. | 환경을 파괴하는 정책이 강력히 반대하다. | 환경을 파괴하는 정책은 시민 단체의 강력한 반발에 부딪힌다. | 시민 단체 환경을 파괴하는 정책을 적극 반대만하고 실천은 하지 않는다. | 3 |
만지다#다래끼#더럽다 | 다래끼가 있는 사람이 만지면 더럽다. | 더러운 사람은 다래끼가 있으니까 만지면 안된다. | 다래끼가 생기면 흙이 묻은 손으로 눈을 만지면 된다. | 더러운 손으로 만지면 다래끼가 생겨버릴거야. | 4 |
그#작전#소대#명령하다#결정하다#지휘 | 소대장인 그는 작전을 명령하지도 않고 소대 지휘를 결정하지도 않았다. | 소대장인 그는 작전을 명령하고 결정했다. | 그는 작전이 명령하고, 소대 지휘를 결정해야 함을 깨닫았다. | 작전을 명령하면서 소대 그에게 지휘가 결정되었다. | 1 |
🌠 KoCommonGEN v2
KoCommonGEN v2: A Benchmark for Navigating Korean Commonsense Reasoning Challenges in Large Language Models (ACL 2024-Findings)
Jaehyung Seo, Jaewook Lee, Chanjun Park, SeongTae Hong, Seungjun Lee and Heuiseok Lim
🏫 NLP & AI Lab, Korea University
🔥 News
- September 27, 2023: Provided data support for the Open Ko-LLM Leaderboard
- August 7, 2024: Dataset Release
- August 10, 2024: Experimental Results for the New Models Added
- August 14, 2024: Presented a research paper at ACL 2024
👥 Human Evaluation
We recruited 22 native Korean speaking volunteers as human evaluators and paid them $0.8 per question.
Model | # | Average Score | cohen's kappa | Krippendorff's alpha |
---|---|---|---|---|
Human | 22 | 0.8395 | 0.7693 | 0.7706 |
🤖 Models (August 10, 2024)
The results of 2-shot evaluation of the newly released models.
Model | Size | Acc_norm | Stderr | Link |
---|---|---|---|---|
GPT-4 (June 13, 2023) | 0.7450 | |||
Mistral-Nemo-Instruct | 12B | 0.6612 | 0.0163 | 🔗 |
Mistral-Nemo-Base | 12B | 0.6340 | 0.0166 | 🔗 |
Meta-Llama-3.1-8B | 8B | 0.6246 | 0.0166 | 🔗 |
QWEN2-7B base | 7B | 0.6187 | 0.0167 | 🔗 |
EXAONE-3.0-7.8B-Instruct | 7.8B | 0.6088 | 0.0168 | 🔗 |
MLP-KTLim-Bllossom-8B | 8B | 0.6057 | 0.0168 | 🔗 |
Meta-Llama-3.1-8B-Instruct | 8B | 0.6057 | 0.0168 | 🔗 |
KULLM3 | 10.8B | 0.6033 | 0.0168 | 🔗 |
QWEN2-7B inst | 7B | 0.5832 | 0.017 | 🔗 |
Gemma-2-9b-it | 9B | 0.5714 | 0.0170 | 🔗 |
Aya-23-8B | 8B | 0.5159 | 0.0172 | 🔗 |
Allganize-Alpha-Instruct | 8B | 0.4970 | 0.0172 | 🔗 |
As mentioned in the paper, it is possible to evaluate various models.
🇰🇷🇺🇸🇯🇵🇨🇳🇪🇸 Code-switching
The dataset can be found on HF中国镜像站 at: nlpai-lab/ko_commongen_v2_code_switching
This dataset contains code-switching data for the following languages:
- Korean (korean)
- English (english)
- Japanese (japan)
- Chinese (china)
- Spanish (espanol)
(The code-switching data relies on machine translation, which may result in some inaccuracies.)
📖 Citation
@inproceedings{seo2024Kocommongenv2,
title = "KoCommonGEN v2: A Benchmark for Navigating Korean Commonsense Reasoning Challenges in Large Language Models",
author = "Jaehyung Seo and Jaewook Lee and Chanjun Park and SeongTae Hong and Seungjun Lee and Heuiseok Lim",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2024",
month = August,
year = "2024",
address = "Bangkok, Thailand",
publisher = "Association for Computational Linguistics",
url = "TBD",
doi = "TBD",
pages = "TBD"}
🚨 Warning!
This dataset contains some instances of toxic speech.
🙏 Acknowledgement
We sincerely appreciate the dedication of Chanjun Park, Sanghoon Kim and Sunghun Kim (Sung Kim) from Upstage AI in managing one of the benchmark datasets for the Open Ko-LLM LeaderBoard.
- Downloads last month
- 133