Instructions to use hell0ks/ja-ko-vn-12b-v2-gguf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- llama-cpp-python
How to use hell0ks/ja-ko-vn-12b-v2-gguf with llama-cpp-python:
# !pip install llama-cpp-python from llama_cpp import Llama llm = Llama.from_pretrained( repo_id="hell0ks/ja-ko-vn-12b-v2-gguf", filename="model-BF16-00001-of-00003.gguf", )
llm.create_chat_completion( messages = "\"ะะตะฝั ะทะพะฒัั ะะพะปััะณะฐะฝะณ ะธ ั ะถะธะฒั ะฒ ะะตัะปะธะฝะต\"" )
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use hell0ks/ja-ko-vn-12b-v2-gguf with llama.cpp:
Install (macOS, Linux)
curl -LsSf https://llama.app/install.sh | sh # Start a local OpenAI-compatible server with a web UI: llama serve -hf hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M # Run inference directly in the terminal: llama cli -hf hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M # Run inference directly in the terminal: llama cli -hf hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M
Use pre-built binary
# Download pre-built binary from: # https://github.com/ggerganov/llama.cpp/releases # Start a local OpenAI-compatible server with a web UI: ./llama-server -hf hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M # Run inference directly in the terminal: ./llama-cli -hf hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M
Build from source code
git clone https://github.com/ggerganov/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-server llama-cli # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama-server -hf hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama-cli -hf hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M
Use Docker
docker model run hf.co/hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M
- LM Studio
- Jan
- Ollama
How to use hell0ks/ja-ko-vn-12b-v2-gguf with Ollama:
ollama run hf.co/hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M
- Unsloth Studio
How to use hell0ks/ja-ko-vn-12b-v2-gguf with Unsloth Studio:
Install Unsloth Studio (macOS, Linux, WSL)
curl -fsSL https://unsloth.ai/install.sh | sh # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for hell0ks/ja-ko-vn-12b-v2-gguf to start chatting
Install Unsloth Studio (Windows)
irm https://unsloth.ai/install.ps1 | iex # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for hell0ks/ja-ko-vn-12b-v2-gguf to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for hell0ks/ja-ko-vn-12b-v2-gguf to start chatting
- Atomic Chat new
- Docker Model Runner
How to use hell0ks/ja-ko-vn-12b-v2-gguf with Docker Model Runner:
docker model run hf.co/hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M
- Lemonade
How to use hell0ks/ja-ko-vn-12b-v2-gguf with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull hell0ks/ja-ko-vn-12b-v2-gguf:Q4_K_M
Run and chat with the model
lemonade run user.ja-ko-vn-12b-v2-gguf-Q4_K_M
List all available models
lemonade list
Run and chat with the model
lemonade run user.ja-ko-vn-12b-v2-gguf-List all available models
lemonade listja-ko-vn-12b V2 (GGUF)
This model translates text from Japanese visual novels or various games into Korean.
์ด ๋ชจ๋ธ์ ์ผ๋ณธ์ด๋ก ๋ ๋น์ฃผ์ผ ๋ ธ๋ฒจ ํน์ ๋ค์ํ ๊ฒ์์ ํ ์คํธ๋ฅผ ํ๊ตญ์ด๋ก ๋ฒ์ญํฉ๋๋ค.
Updates
- 2025/12/13 - Quant upload
- 2025/12/13 - ์์ํ ์ ๋ก๋
Model Details
Model Description
Google์ Gemma 3 12B ๋ชจ๋ธ์ ๊ธฐ๋ฐ์ผ๋ก ํ์ฌ ๋น์ฃผ์ผ ๋ ธ๋ฒจ ๋ฑ ๊ฒ์๋ฅ์ ์ ํฉํ ๋ฒ์ญ์ ํ๋๋ก ๋ค์ ์ ์ฐจ๋ฅผ ๊ฑฐ์ณค์ต๋๋ค:
- ์ผ๋ณธ์ด, ํ๊ตญ์ด์ ๋ํ ๋ฌธํ ๊ณ์ด ๋๋ฉ์ธ ์ธ์ด ํ์ต (CPT)
- ๋น์ฃผ์ผ ๋ ธ๋ฒจ์ ์๋ฌธ๊ณผ ๋ฒ์ญ๋ฌธ ํ์ต (SFT)
- ํน์ ์ํฉ ๋ฐ ๋จ์ด์ ๋ํ ๋๋ฉ์ธ ํ์ต (DPO)
์ฃผ๋ก "์ง์ญ"์ ์ ํธํ์๋ ๋ถ๊ป ์ ํฉํฉ๋๋ค.
- Developed by: hell0ks
- Model type: Translation
- Language(s) (NLP): Japanese(Input), Korean(Output)
- License: Gemma
- Finetuned from model : google/gemma-3-12b-pt
- Max context length : 4096 (๊น์ง๋ง ํ ์คํธ ๋จ)
V1๊ณผ์ ์ฐจ์ด์
- Base model์ด Tri-7B์์ Gemma-3-12B๋ก ๋ณ๊ฒฝ๋์์ต๋๋ค.
- CPT, DPO ํ์ต์ ํตํด ์ข ๋ ์์ฐ์ค๋ฌ์ด ์ดํ๋ฅผ ์ฌ์ฉํ๋๋ก ์ ๋ํ์ต๋๋ค.
- ๊ณ ์ ๋ช ์ฌ ๊ณ ์ ๊ธฐ๋ฅ์ด ๋ถ์์ ํ ์ํ๋ก ๋ฆด๋ฆฌ์ฆ ๋ ๊ฒ์ ์์ ํ์ต๋๋ค.
Model Sources
- Repository: hell0ks/ja-ko-vn-12b-v2
Uses
- Temperature: 0.1, Top_k = 0.95, repetition_penalty 1.05 ~ 1.1๋ฅผ ์ถ์ฒ๋๋ฆฝ๋๋ค.
- ๊ณ ์ ๋ช
์ฌ, ์ด๋ฆ ๋ฑ์ System prompt๋ก ํํธ๋ฅผ ์ฃผ์ค ์ ์์ต๋๋ค. ์:
ๅฒก้จๅซๅคช้=์ค์นด๋ฒ ๋ฆฐํ๋ก,้ฟไธ้ณ้ด็พฝ=์๋ง๋ค ์ค์ฆํ - ํ๋กฌํํธ๋ ์ผ๋ณธ์ด ์๋ฌธ๋ง ์ ๋ ฅํ์ธ์. ์ ์ด ์ฝ๋๋ ํน์๋ฌธ์๋ ์ต๋ํ ๊ทธ๋๋ก ์ ์งํ๋๋ก ํ์ต๋์์ต๋๋ค.
- Chat ๋ชจ๋๋ก ์ฌ์ฉํ์ง ๋ง์๊ณ , Completions ๋ชจ๋๋ก ์ฌ์ฉํ์ธ์. ์ฑ๊ธ ํด์ผ๋ก๋ง ํ์ต๋์์ต๋๋ค.
- llama.cpp์์ ์ฌ์ฉํ์ค ๋ ๊ผญ --jinja ํ๋๊ทธ๋ฅผ ์ฌ์ฉํ์ธ์.
Out-of-Scope Use
- ๋ฒ์ญ์ ์ ํ์ฑ์ด ํฌ๊ฒ ์ค์ํ์ง ์์ ์์ ์๋ง ์ฌ์ฉํ์ธ์.
Bias, Risks, and Limitations
- ์ผ๋ณธ์ด์์ ํ๊ตญ์ด ๋ฒ์ญ๋ง ์ง์ํฉ๋๋ค. ๋ฐ๋๋ ๋ค๋ฅธ ์ธ์ด๋ ์ง์ํ์ง ์์ต๋๋ค.
- Safety RL์ด ๋์ง ์์์ต๋๋ค. ์ ์ ๊ฐ ์ ๋ ฅํ ๋์ฌ์ ๋ํด ๊ทธ๋๋ก ๋ฒ์ญํฉ๋๋ค.
Recommendations
- ์ฌ์ฉ์(์ง์ ๋ฐ ํ์ ์ฌ์ฉ์ ๋ชจ๋)๋ ๋ชจ๋ธ์ ์ํ์ฑ, ํธํฅ์ฑ ๋ฐ ํ๊ณ์ ์ ์ธ์งํด์ผ ํฉ๋๋ค.
- ๊ธด ํ ์คํธ๋ฅผ ๋ฒ์ญํด์ผ ํ๋ ๊ฒฝ์ฐ ๋ฌธ๋จ ํน์ ๋ฌธ์ฅ ๋จ์๋ก ์ด์ฉํ์๋ ๊ฒ์ ์ถ์ฒ๋๋ฆฝ๋๋ค.
Technical Specifications
Hardware
Nvidia DGX Spark, 2 Nodes
Software
Axolotl
Acknowledgement
ํ์ต ํ๋์จ์ด๋ฅผ ๋๊ฐ ์์ด ์ง์ํด์ฃผ์ ์ต๋ช ์ ๋ถ๊ป ๊ฐ์ฌ ์ธ์ฌ ๋๋ฆฝ๋๋ค.
- Downloads last month
- 323
3-bit
4-bit
5-bit
6-bit
8-bit
16-bit
Pull the model
# Download Lemonade from https://lemonade-server.ai/