README.md

---
library_name: transformers
tags:
- pytorch
license: llama3
language:
- ko
pipeline_tag: text-generation
---

<p align="left">
  <img src="https://huggingface.co/cpm-ai/Ocelot-Ko-self-instruction-10.8B-v1.0/resolve/main/ocelot.webp" width="50%"/>
<p>
  
# solar-kor-resume
> Update @ 2024.06.05: First release of Llama3-Ocelot-8B-instruct-v01
<!-- Provide a quick summary of what the model is/does. -->

This model card corresponds to the 10.8B Instruct version of the **Llama-Ko** model. 

The train wad done on A100-80GB

**Resources and Technical Documentation**:
* [llama Model](beomi/Llama-3-Open-Ko-8B)
- [Orca-Math](https://huggingface.co/datasets/kuotient/orca-math-korean-dpo-pairs)
- [ko_Ultrafeedback_binarized](maywell/ko_Ultrafeedback_binarized)

**Citation**

**Model Developers**: frcp, nebchi, pepperonipizza97

## Model Information
It is an LLM model capable of generating Korean text, trained on a pre-trained base model with high-quality Korean SFT dataset and DPO dataset.

#### *Inputs and outputs*

- **Input:** Text string, such as a question, a prompt, or a document to be summarized.
- **Output:** Generated Korean-language text in response to the input, such as an answer to a question, or a summary of a document.

#### Running the model on a single / multi GPU
```python
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("cpm-ai/Ocelot-Ko-self-instruction-10.8B-v1.0")
model = AutoModelForCausalLM.from_pretrained("cpm-ai/Ocelot-Ko-self-instruction-10.8B-v1.0", device_map="auto")

pipe = pipeline("text-generation", model=model, tokenizer=tokenizer, max_new_tokens=4096, streamer=streamer)

text = '대한민국의 수도는 어디인가요?'

messages = [
    {
        "role": "user",
        "content": "{}".format(text)
    }
]

prompt = pipe.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)

outputs = pipe(
    prompt,
    temperature=0.2,
    add_special_tokens=True
)
print(outputs[0]["generated_text"][len(prompt):])

```
### results

```python
대한민국의 수도는 서울특별시입니다.
서울특별시에는 청와대, 국회의사당, 대법원 등 대한민국의 주요 정부기관이 위치해 있습니다.
또한 서울시는 대한민국의 경제, 문화, 교육, 교통의 중심지로써 대한민국의 수도이자 대표 도시입니다.제가 도움이 되었길 바랍니다. 더 궁금한 점이 있으시면 언제든지 물어보세요!
```

```bibtex
@misc {cpm-ai/Ocelot-Ko-self-instruction-10.8B-v1.0,
	author       = { {frcp, nebchi, pepperonipizza97} },
	title        = { solar-kor-resume},
	year         = 2024,
	url          = { https://huggingface.co/cpm-ai/Ocelot-Ko-self-instruction-10.8B-v1.0 },
	publisher    = { Hugging Face }
}
```

Results in [LogicKor](https://github.com/StableFluffy/LogicKor)* are as follows:

|         Model                  | Single turn* | Multi turn* | Overall* |
|:------------------------------:|:------------:|:-----------:|:--------:|
| gemini-1.5-pro-preview-0215 | 7.90 | 6.26 | 7.08 |
| xionic-1-72b-20240404 | 7.23 | 6.28 | 6.76 |
| Ocelot-Instruct  |     6.79    |    **6.71**    |   6.75  |
| allganize/Llama-3-Alpha-Ko-8B-Instruct | 7.14 | 6.09 | 6.61 |