README.md

---
language:
- en
- ko
datasets:
- kyujinpy/KOpen-platypus
library_name: transformers
pipeline_tag: text-generation
license: cc-by-nc-sa-4.0
---
**(주)미디어그룹사람과숲과 (주)마커의 LLM 연구 컨소시엄에서 개발된 모델입니다**  
**The license is `cc-by-nc-sa-4.0`.**  

# **Ko-Platypus2-13B**  
![KO-Platypus2-13B](./KO_platypus.png)


## Model Details

**More detail repo(Github): [KO-Platypus](https://github.com/Marker-Inc-Korea/KO-Platypus)**

**Model Developers** Kyujin Han (kyujinpy)

**Input** Models input text only.

**Output** Models generate text only.

**Model Architecture** 
KO-Platypus2-13B is an auto-regressive language model based on the LLaMA2 transformer architecture.

**Base Model** [hyunseoki/ko-en-llama2-13b](https://huggingface.co/hyunseoki/ko-en-llama2-13b)  

**Training Dataset**
I use [KOpen-platypus](https://huggingface.co/datasets/kyujinpy/KOpen-platypus).  
It is high-quality korean translation dataset about [open-platypus](https://huggingface.co/datasets/garage-bAInd/Open-Platypus).  

I use A100 GPU 40GB and COLAB, when trianing.  

# **Model Benchmark**

## KO-LLM leaderboard
- Follow up as [Open KO-LLM LeaderBoard](https://huggingface.co/spaces/upstage/open-ko-llm-leaderboard).  
  
![img](./leaderboard.png)
| Model | Average |Ko-ARC | Ko-HellaSwag | Ko-MMLU | Ko-TruthfulQA | Ko-CommonGen V2 |
| --- | --- | --- | --- | --- | --- | --- |
| KO-Platypus2-13B(ours) | 47.90 | 44.20 | 54.31 | 42.47 | 44.41 | 54.11 | 
| [hyunseoki/ko-en-llama2-13b](https://huggingface.co/hyunseoki/ko-en-llama2-13b) | 46.68 | 42.15 | 54.23 | 38.90 | 40.74 | 57.39 |
| [MarkrAI/kyujin-CoTy-platypus-ko-12.8b](https://huggingface.co/MarkrAI/kyujin-CoTy-platypus-ko-12.8b) | 46.44 | 34.98 | 49.11 | 25.68 | 37.59 | 84.86 | 
| [momo/polyglot-ko-12.8b-Chat-QLoRA-Merge](https://huggingface.co/momo/polyglot-ko-12.8b-Chat-QLoRA-Merge) | 45.71 | 35.49 | 49.93 | 25.97 | 39.43 | 77.70 |   
| [KoT-platypus2-7B](https://huggingface.co/kyujinpy/KoT-platypus2-7B) | 45.62 | 38.05 | 49.63 | 34.68 | 37.69 | 68.08 |  
> Compare with Top 4 SOTA models. (update: 10/06)

---  
# Implementation Code
```python
### KO-Platypus
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

repo = "kyujinpy/KO-Platypus2-13B"
CoT-llama = AutoModelForCausalLM.from_pretrained(
        repo,
        return_dict=True,
        torch_dtype=torch.float16,
        device_map='auto'
)
CoT-llama_tokenizer = AutoTokenizer.from_pretrained(repo)
```

> Readme format: [kyujinpy/KoT-platypus2-7B](https://huggingface.co/kyujinpy/KoT-platypus2-7B)

---