Text Generation
Transformers
Safetensors
Korean
English
gemma3n
image-text-to-text
gemma
gemma-3n
korean
qlora
instruction-tuning
conversational
Instructions to use ION-Communications/gemma-3n-E2B-ko with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use ION-Communications/gemma-3n-E2B-ko with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="ION-Communications/gemma-3n-E2B-ko") 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, AutoModelForImageTextToText processor = AutoProcessor.from_pretrained("ION-Communications/gemma-3n-E2B-ko") model = AutoModelForImageTextToText.from_pretrained("ION-Communications/gemma-3n-E2B-ko") 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 ION-Communications/gemma-3n-E2B-ko with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "ION-Communications/gemma-3n-E2B-ko" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "ION-Communications/gemma-3n-E2B-ko", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/ION-Communications/gemma-3n-E2B-ko
- SGLang
How to use ION-Communications/gemma-3n-E2B-ko 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 "ION-Communications/gemma-3n-E2B-ko" \ --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": "ION-Communications/gemma-3n-E2B-ko", "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 "ION-Communications/gemma-3n-E2B-ko" \ --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": "ION-Communications/gemma-3n-E2B-ko", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use ION-Communications/gemma-3n-E2B-ko with Docker Model Runner:
docker model run hf.co/ION-Communications/gemma-3n-E2B-ko
gemma-3n-E2B-ko (QLoRA fine-tuned)
google/gemma-3n-E2B-it ๋ฅผ ํ๊ตญ์ด instruction ๋ฐ์ดํฐ(KoAlpaca v1.1a)๋ก QLoRA ์ถ๊ฐ ํ์ตํ ๋ชจ๋ธ์
๋๋ค.
ํ์ต ๊ฐ์
- ๋ฒ ์ด์ค ๋ชจ๋ธ: google/gemma-3n-E2B-it (์ ํจ ํ๋ผ๋ฏธํฐ ~2B)
- ๋ฐฉ๋ฒ: QLoRA (4-bit, LoRA rank=16, alpha=16), Unsloth
- ๋ฐ์ดํฐ: beomi/KoAlpaca-v1.1a (ํ๊ตญ์ด instruction)
- ํ๋์จ์ด: ๋จ์ผ NVIDIA H100
- ์ต๋ ์ํ์ค ๊ธธ์ด: 1024
- ํ์ต ์คํ : 500 steps (effective batch size 8, learning rate 2e-4, linear schedule)
- LoRA ํ์ต ํ๋ผ๋ฏธํฐ: 21M / 5.46B (0.39%)
- ์ต์ข train loss: ์ฝ 2.33
ํ์ ๋ผ์ด๋ธ๋ฌ๋ฆฌ ์ค์น
- pip install timm pillow accelerate torch transformers
์ฌ์ฉ ์์
from transformers import AutoProcessor, AutoModelForImageTextToText
import torch
model_id = "ION-Communications/gemma-3n-E2B-ko"
processor = AutoProcessor.from_pretrained(model_id)
model = AutoModelForImageTextToText.from_pretrained(
model_id,
dtype=torch.bfloat16,
device_map="auto",
)
messages = [
{
"role": "user",
"content": [
{"type": "text", "text": "ํ๊ตญ์ ์๋๋?"}
],
}
]
input_ids = processor.apply_chat_template(
messages,
add_generation_prompt=True,
tokenize=True,
return_tensors="pt",
).to(model.device)
out = model.generate(
input_ids=input_ids,
max_new_tokens=128,
)
print(processor.decode(out[0], skip_special_tokens=True))
ํ๊ณ
- ์๊ท๋ชจ instruction ๋ฐ์ดํฐ๋ก ์งง๊ฒ ์ถ๊ฐ ํ์ตํ ๋ชจ๋ธ๋ก, ์ฌ์ค์ฑ/์์ ์ฑ ๋ณด์ฅ์ ์ ํ์ ์ ๋๋ค.
- ๋ฒ ์ด์ค Gemma 3n ๋ผ์ด์ ์ค(Gemma Terms of Use)๋ฅผ ๋ฐ๋ฆ ๋๋ค.
- Downloads last month
- 27