Instructions to use hyunjae/skt-kogpt2-kullm-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hyunjae/skt-kogpt2-kullm-v2 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="hyunjae/skt-kogpt2-kullm-v2")

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("hyunjae/skt-kogpt2-kullm-v2")
model = AutoModelForMultimodalLM.from_pretrained("hyunjae/skt-kogpt2-kullm-v2")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use hyunjae/skt-kogpt2-kullm-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hyunjae/skt-kogpt2-kullm-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hyunjae/skt-kogpt2-kullm-v2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/hyunjae/skt-kogpt2-kullm-v2

SGLang

How to use hyunjae/skt-kogpt2-kullm-v2 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "hyunjae/skt-kogpt2-kullm-v2" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hyunjae/skt-kogpt2-kullm-v2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "hyunjae/skt-kogpt2-kullm-v2" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hyunjae/skt-kogpt2-kullm-v2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use hyunjae/skt-kogpt2-kullm-v2 with Docker Model Runner:
```
docker model run hf.co/hyunjae/skt-kogpt2-kullm-v2
```

교육용으로 학습 한 간단한 instruction fine-tuning 모델 (updated 2023/08/06)

Pretrained model: skt/kogpt2-base-v2 (https://github.com/SKT-AI/KoGPT2)
Training data: kullm-v2(https://huggingface.co/datasets/nlpai-lab/kullm-v2)

from transformers import AutoModelForCausalLM
from transformers import PreTrainedTokenizerFast

tokenizer = PreTrainedTokenizerFast.from_pretrained("hyunjae/skt-kogpt2-kullm-v2",
                                                    bos_token='</s>', eos_token='</s>', unk_token='<unk>',
                                                    pad_token='<pad>', mask_token='<mask>', padding_side="right", model_max_length=512)
model = AutoModelForCausalLM.from_pretrained('hyunjae/skt-kogpt2-kullm-v2').to('cuda')

PROMPT= "### system:사용자의 질문에 맞는 적절한 응답을 생성하세요.\n### 사용자:{instruction}\n### 응답:"
text = PROMPT.format_map({'instruction':"안녕? 너가 할 수 있는게 뭐야?"})
input_ids = tokenizer.encode(text, return_tensors='pt').to(model.device)

gen_ids = model.generate(input_ids,
                        repetition_penalty=2.0,
                        pad_token_id=tokenizer.pad_token_id,
                        eos_token_id=tokenizer.eos_token_id,
                        bos_token_id=tokenizer.bos_token_id,
                        num_beams=4,
                        no_repeat_ngram_size=4,
                        max_new_tokens=128,
                        do_sample=True,
                        top_k=50)


generated = tokenizer.decode(gen_ids[0])
print(generated)

Downloads last month: 4

hyunjae
/

skt-kogpt2-kullm-v2

Dataset used to train hyunjae/skt-kogpt2-kullm-v2