Instructions to use Bllossom/llama-3.2-Korean-Bllossom-AICA-5B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Bllossom/llama-3.2-Korean-Bllossom-AICA-5B with Transformers:

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

pipe = pipeline("image-text-to-text", model="Bllossom/llama-3.2-Korean-Bllossom-AICA-5B")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("Bllossom/llama-3.2-Korean-Bllossom-AICA-5B")
model = AutoModelForMultimodalLM.from_pretrained("Bllossom/llama-3.2-Korean-Bllossom-AICA-5B")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Bllossom/llama-3.2-Korean-Bllossom-AICA-5B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Bllossom/llama-3.2-Korean-Bllossom-AICA-5B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Bllossom/llama-3.2-Korean-Bllossom-AICA-5B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/Bllossom/llama-3.2-Korean-Bllossom-AICA-5B

SGLang

How to use Bllossom/llama-3.2-Korean-Bllossom-AICA-5B 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 "Bllossom/llama-3.2-Korean-Bllossom-AICA-5B" \
    --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": "Bllossom/llama-3.2-Korean-Bllossom-AICA-5B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "Bllossom/llama-3.2-Korean-Bllossom-AICA-5B" \
        --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": "Bllossom/llama-3.2-Korean-Bllossom-AICA-5B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use Bllossom/llama-3.2-Korean-Bllossom-AICA-5B with Docker Model Runner:
```
docker model run hf.co/Bllossom/llama-3.2-Korean-Bllossom-AICA-5B
```

prompt 문의

by HugJihun - opened Mar 14, 2025

Discussion

HugJihun

Mar 14, 2025

안녕하세요!
Bllossom/llama-3.2-Korean-Bllossom-AICA-5B 사용해서 local ai 개발중인데
prompt 적용은 어떻게 하는지 알 수 있을까요?
chat_template.json 에서 system 부분을 수정해도 수정한 내용대로 model 이 적용이 안 됩니다,,
chat_template.json 에서 prompt 조정하는게 맞을까요?

PerRing

Bllossom org Mar 14, 2025

MllamaProcessor는 image가 들어갈 경우 시스템 메시지가 출력하지 않도록 구성되어 있습니다.
image를 사용하지 않고 text만 사용할경우 chat_template.json 에서 system 부분을 수정하면 정상적으로 작동하시는 걸 볼 수 있습니다.

HugJihun

Mar 14, 2025

답변 감사합니다! image 사용하지 않고 text만 사용하니 chat_template.json 에서 system 부분 적용이 됩니다! 감사합니다.
prompt 를 꼭 chat_template.json 에서 사용을 해야 되는건가요? 내부소스에 prompt를 만들어서 적용을 할 순 없을까요?

PerRing

Bllossom org Mar 14, 2025

chat_template.json을 직접 수정하지 않고 아래의 코드와 같이 코드상에서 프로세서를 불러온 뒤 chat_template을 지정해서 사용하실수 있습니다

processor = MllamaProcessor.from_pretrained('Bllossom/llama-3.2-Korean-Bllossom-AICA-5B')
my_template="""
{%- for message in messages %}
  {%- if message['role'] == 'user' %}
    {{ bos_token + '[USER] ' + message['content'].strip() + ' [/USER]' }}
  {%- elif message['role'] == 'assistant' %}
    {{ '[ASST] ' + message['content'] + ' [/ASST]' + eos_token }}
  {%- endif %}
{%- endfor %}"""
processor.chat_template = my_template

jinja2 template에 대한 간단한 예제들은 아래의 링크에서 확인하시길바랍니다!
https://huggingface.co/docs/transformers/main/en/chat_templating

HugJihun

Mar 17, 2025

항상 친절하고 자세한 답변 감사합니다!
혹시 이전 대화 내용을 저장하는 것에 대하여 조언을 얻을 수 있을까요?
이전 대화 내용을 DB에 저장하고 대화 내용을 통째로 받아서 사용해봤는데, 대화 내용이 커질수록 토큰값 문제인건지 답변이 제대로 안 나오더라고요,,
대화 내용을 어떻게 저장하고 사용하는지 궁금합니다!

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment