README.md

---
license: apache-2.0
tags:
- generated_from_trainer
base_model: yanolja/EEVE-Korean-2.8B-v1.0

---
<p align="left">
  <img src="https://huggingface.co/crimsonjoo/Neversleep-3B-Instruct-v0.1/resolve/main/neversleep_logo.webp" width="100%"/>
<p>

# "We must sleep, but AI Never Sleeps!"
&nbsp;

## Prompt Template
```
A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.
Human: {prompt}
Assistant:
```
## Simple-Usage
```python
from transformers import AutoTokenizer
from transformers import AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained("yanolja/EEVE-Korean-Instruct-2.8B-v1.0", trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained("yanolja/EEVE-Korean-Instruct-2.8B-v1.0", trust_remote_code=True)

prompt_template = "A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.\nHuman: {prompt}\nAssistant:\n"
text = '다이어트식 메뉴를 추천해주세요.\n\n(A) 샐러드\n(B) 치킨\n(C) 피자\n(D) 파스타'
model_inputs = tokenizer(prompt_template.format(prompt=text), return_tensors='pt')

outputs = model.generate(**model_inputs, max_new_tokens=256)
output_text = tokenizer.batch_decode(outputs, skip_special_tokens=True)[0]
print(output_text)
```

### Example Output
```
A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.
Human: 다이어트식 메뉴를 추천해주세요.

(A) 샐러드
(B) 치킨
(C) 피자
(D) 파스타
Assistant:
(A) 샐러드를 추천드립니다. 샐러드는 저칼로리이면서도 영양소가 풍부해 다이어트식으로 적합합니다. 다양한 채소와 단백질을 추가하여 균형 잡힌 식사를 만드실 수 있습니다.
```

## About the Model

First of all, Overwhelming gratitude to 'yanolja/EEVE' Model & Team! 
This model is a fine-tuned version of [crimsonjoo/Neversleep-3B-v0.1](https://huggingface.co/crimsonjoo/Neversleep-3B-v0.1), which is a Korean vocabulary-extended version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2). Specifically, we utilized Direct Preference Optimization (DPO) through the use of [Axolotl](https://github.com/OpenAccess-AI-Collective/axolotl).

For more details, please refer to our technical report: [Efficient and Effective Vocabulary Expansion Towards Multilingual Large Language Models](https://arxiv.org/abs/2402.14714).

## Training Data
  - Korean-translated version of [Open-Orca/SlimOrca-Dedup](https://huggingface.co/datasets/Open-Orca/SlimOrca-Dedup)
  - Korean-translated version of [argilla/ultrafeedback-binarized-preferences-cleaned](https://huggingface.co/datasets/argilla/ultrafeedback-binarized-preferences-cleaned)
  - No other dataset was used