Instructions to use javiagu/KULLM3_GPTQ with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use javiagu/KULLM3_GPTQ with Transformers:

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

pipe = pipeline("text-generation", model="javiagu/KULLM3_GPTQ")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("javiagu/KULLM3_GPTQ")
model = AutoModelForCausalLM.from_pretrained("javiagu/KULLM3_GPTQ")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use javiagu/KULLM3_GPTQ with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "javiagu/KULLM3_GPTQ"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "javiagu/KULLM3_GPTQ",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/javiagu/KULLM3_GPTQ

SGLang

How to use javiagu/KULLM3_GPTQ 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 "javiagu/KULLM3_GPTQ" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "javiagu/KULLM3_GPTQ",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "javiagu/KULLM3_GPTQ" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "javiagu/KULLM3_GPTQ",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use javiagu/KULLM3_GPTQ with Docker Model Runner:
```
docker model run hf.co/javiagu/KULLM3_GPTQ
```

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

GPTQ Quantized model of KULLM3 Korea University NLP AI Lab.

I did the conversion to GPTQ, the whole model was built by NLP AI Lab, therefore, all my credits to them.

The model seems to work really well in GPTQ and it seems a new step towards a fully usable korean LLM.

It was converted using 500 samples of korean wiki and 500 samples of english wiki.

Amazing work!

The original repo: https://huggingface.co/nlpai-lab/KULLM3

Downloads last month: 1

Safetensors

Model size

11B params

Tensor type

I32

F16