Transformers
GGUF
Merge
How to use from
llama.cpp
Install from brew
brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf muzerai/Llama-3.1-KoEn-8B-magic8-GGUF:
# Run inference directly in the terminal:
llama-cli -hf muzerai/Llama-3.1-KoEn-8B-magic8-GGUF:
Install from WinGet (Windows)
winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf muzerai/Llama-3.1-KoEn-8B-magic8-GGUF:
# Run inference directly in the terminal:
llama-cli -hf muzerai/Llama-3.1-KoEn-8B-magic8-GGUF:
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 muzerai/Llama-3.1-KoEn-8B-magic8-GGUF:
# Run inference directly in the terminal:
./llama-cli -hf muzerai/Llama-3.1-KoEn-8B-magic8-GGUF:
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 muzerai/Llama-3.1-KoEn-8B-magic8-GGUF:
# Run inference directly in the terminal:
./build/bin/llama-cli -hf muzerai/Llama-3.1-KoEn-8B-magic8-GGUF:
Use Docker
docker model run hf.co/muzerai/Llama-3.1-KoEn-8B-magic8-GGUF:
Quick Links

Llama-3.1-Ko-8B-magic8 made by "AIJOAH"

The merged model combining Llama-3.1-Korean-8B-Instruct and Llama-3.1-Storm-8B improves performance like โ€” including Korean-language instruction following, multilingual knowledge-based QA, reasoning, reduced hallucinations, and structured output generation (e.g., JSON, Markdown). This merge is particularly beneficial for developers seeking a strong Korean-capable model that also excels in logic, accuracy, and function-calling, while remaining lightweight enough for local inference environments such as Ollama or vLLM.

Merge Method

This model was merged using the DELLA

Models Merged

The following models were included in the merge:

Test Q5

ollama create modelname -f Modelfile
>>> ์ด๋ ฅ์„œ ์“ฐ๋Š” ๋ฐฉ๋ฒ•์„ ์„ค๋ช…ํ•ด์ค˜
์ด๋ ฅ์„œ ๋˜๋Š” ๋ฆฌ์กธ๋ฃจ์…˜(Resume)๋ž€, ์ž์‹ ์˜ ๊ฒฝ๋ ฅ์„ ์š”์•ฝํ•˜์—ฌ ์ •๋ฆฌํ•œ ๋ฌธ์„œ๋กœ, ์ „ ์ง๋ฌด๋ฅผ ํฌํ•จํ•œ ๊ฒฝํ—˜๊ณผ ์ž๊ฒฉ์„ ๋ช…ํ™•ํ•˜๊ฒŒ ํ‘œํ˜„ํ•˜๋Š” ๋ฐ ์‚ฌ์šฉ๋ฉ๋‹ˆ๋‹ค. ์ด๋ ฅ์„œ๋Š” ์ฃผ์š” ๊ธฐ์—…์ด๋‚˜ ์กฐ์ง์ด ์ง€์›
์ž๋ฅผ ํ‰๊ฐ€ํ•˜๊ณ  ์ตœ์ข…์ ์œผ๋กœ ์ตœ์ ์˜ ์ธ์žฌ๋ฅผ ์„ ๋ฐœํ•˜๊ธฐ ์œ„ํ•ด ํ•„์ˆ˜์ ์ธ ๋„๊ตฌ์ž…๋‹ˆ๋‹ค.

1. ์ด๋ ฅ์„œ ์ž‘์„ฑ์— ํ•„์š”ํ•œ ์ •๋ณด:

- ์ด๋ฆ„ ๋ฐ ์—ฐ๋ฝ์ฒ˜
- ์ง์—… ์ œ๋ชฉ ๋ฐ ์ง์œ„
- ํ˜„์žฌ/์ „ ์ง๋ฌด ๋ฐ ํšŒ์‚ฌ๋ช… (์ตœ๊ทผ 10๋…„)
- ๊ต์œก: ํ•™๋ถ€/๋Œ€ํ•™, ๋Œ€ํ•™์›, ํŠนํ—ˆ ๋“ฑ
- ๊ธฐ์ˆ ์Šคํ‚ฌ:
  - ํ”„๋กœ๊ทธ๋ž˜๋ฐ ์–ธ์–ด: Java, Python, C++, JavaScript, R etc.
  - ๋ฐ์ดํ„ฐ ๋ถ„์„: R, Python, SQL, Excel ๋“ฑ
  - ๋ฐ์ดํ„ฐ ๋ฒ ์ด์Šค: MySQL, PostgreSQL, MongoDB, Oracle ๋“ฑ
  - ์šด์˜์ฒด์ œ: Windows, Linux, Unix ๋“ฑ

2. ์ด๋ ฅ์„œ์˜ ๋‚ด์šฉ:

- **์‹ ๋ฌธ๊ณ ์ง€**: ์ž์‹ ์˜ ์ด๋ฆ„, ์—ฐ๋ฝ์ฒ˜, ์ง์—… ๋ฐ ํšŒ์‚ฌ๋ช…, ์ง์œ„ ๋“ฑ์„ ๋ณด์—ฌ์ฃผ๋Š” ์˜์—ญ์ž…๋‹ˆ๋‹ค.
- **์ง๋ฌด ๊ฒฝ๋ ฅ**: ์ง€์›์ž๊ฐ€ ๊ฐ€์ง€๊ณ  ์žˆ๋Š” ์ฃผ์š” ๊ฒฝํ—˜๊ณผ ์—ญํ• ์„ ํ‘œํ˜„ํ•ฉ๋‹ˆ๋‹ค. ์ผ๋ฐ˜์ ์œผ๋กœ ๊ฐ€์žฅ ์ตœ๊ทผ๋ถ€ํ„ฐ ์‹œ์ž‘ํ•˜์—ฌ, ์ผ์ž, ํšŒ์‚ฌ๋ช…, ํšŒ์‚ฌ ์œ„์น˜, ์ง์ œ ๋ฐ ์—…๋ฌด ๋‚ด์šฉ์„ ํฌํ•จํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
- **๊ต์œก/ํ›ˆ๋ จ**: ์ง€์›์ž์˜ ํ•™๋ถ€, ๋Œ€ํ•™์›, ํŠนํ—ˆ, ์‹ฌ๋ฆฌํ…Œ์ŠคํŠธ ๋“ฑ์— ๋Œ€ํ•œ ์ •๋ณด๋ฅผ ํฌํ•จํ•ฉ๋‹ˆ๋‹ค.
- **๊ธฐ์ˆ ์Šคํ‚ฌ**: ์ง€์›์ž๊ฐ€ ๊ฐ€์ง€๊ณ  ์žˆ๋Š” ๊ธฐ์ˆ ์Šคํ‚ฌ์— ๋Œ€ํ•ด ๊ฐ„๋‹จํ•˜๊ฒŒ ์„ค๋ช…ํ•ฉ๋‹ˆ๋‹ค. ์ผ๋ฐ˜์ ์œผ๋กœ 3-5๊ฐœ์˜ ๊ด€๋ จ๋œ ๊ธฐ์ˆ ์Šคํ‚ฌ์„ ํฌํ•จํ•ฉ๋‹ˆ๋‹ค.

3. ์ด๋ ฅ์„œ์˜ ํ˜•์‹:

์ด๋ ฅ์„œ์—๋Š” ํ…์ŠคํŠธ, ํ‘œ, ๊ทธ๋ž˜ํ”„ ๋“ฑ ๋‹ค์–‘ํ•œ ๋ฐ์ดํ„ฐ๋ฅผ ํ‘œํ˜„ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ํ•˜์ง€๋งŒ, ์ด๋ ฅ์„œ ์ž‘์„ฑ ์‹œ๋Š” ๋‹ค์Œ๊ณผ ๊ฐ™์€ ์‚ฌํ•ญ์— ์ฃผ์˜ํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
- **๋‹จ๋ฝ**: 1~2์ค„๋กœ ์งง๊ฒŒ ์ž‘์„ฑํ•˜์—ฌ, ๋‚ด์šฉ์„ ์ฝ๊ธฐ ํŽธํ•˜๊ฒŒ ๋งŒ๋“ญ๋‹ˆ๋‹ค.
- **ํ‚ค์›Œ๋“œ**: ์ง€์›์ž๊ฐ€ ๊ฐ€์ง€๊ณ  ์žˆ๋Š” ๊ธฐ์ˆ ์Šคํ‚ฌ ๋ฐ ์ž๊ฒฉ์„ ํ‚ค์›Œ๋“œ๋กœ ํ‘œํ˜„ํ•ฉ๋‹ˆ๋‹ค. ์ด๋ ฅ์„œ๋ฅผ ๊ฒ€์ƒ‰ํ•  ๋•Œ, ์ด๋Ÿฌํ•œ ํ‚ค์›Œ๋“œ๋ฅผ ๋„์›€์ด ๋ฉ๋‹ˆ๋‹ค.
- **ํ˜•์‹**: ์ง€์›์ž๊ฐ€ ๋ณด์œ ํ•œ ๊ฒฝ๋ ฅ์„ ์ˆœ์ฐจ์ ์œผ๋กœ ์ •๋ฆฌํ•˜์—ฌ, ๊ฐ€์žฅ ์ตœ๊ทผ๋ถ€ํ„ฐ ๊ฐ€์žฅ ์˜ค๋ž˜์ „์œผ๋กœ ์ž‘์„ฑํ•ฉ๋‹ˆ๋‹ค.

4. ์ด๋ ฅ์„œ ์ž‘์„ฑ์˜ ์›์น™:

์ด๋ ฅ์„œ๋Š” ์ง€์›์ž์˜ ๊ฒฝํ—˜๊ณผ ๊ธฐ์ˆ ์Šคํ‚ฌ์„ ๋ช…ํ™•ํ•˜๊ณ , ์ผ๊ด€๋˜๊ฒŒ ํ‘œํ˜„ํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค. ์ง€์›์ž๋Š” ๋‹ค์Œ์˜ ์›์น™์— ๋”ฐ๋ผ ์ด๋ ฅ์„œ๋ฅผ ์ž‘์„ฑํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
- **์‚ฌ์‹ค๊ณผ ์‚ฌ์‹ค** : ์ž์‹ ์˜ ๊ฒฝํ—˜๊ณผ ์ž๊ฒฉ์— ๋Œ€ํ•ด ์ง„์‹คํ•˜๊ฒŒ ๊ธฐ์žฌํ•˜์—ฌ์•ผ ํ•ฉ๋‹ˆ๋‹ค.
- **์ผ๊ด€์„ฑ** : ๋™์ผํ•œ ์˜๋ฏธ์™€ ์‚ฌ์šฉ๋ฒ•์„ ์‚ฌ์šฉํ•˜๊ณ , ์ผ๊ด€์„ฑ์„ ์œ ์ง€ํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
- **์‹œ๊ฐ์  ํ‘œํ˜„** : ์ด๋ ฅ์„œ๋ฅผ ํ†ตํ•ด ์ง€์›์ž์˜ ๊ฒฝ๋ ฅ์„ ์‹œ๊ฐ์ ์œผ๋กœ ์‰ฝ๊ฒŒ ํŒŒ์•…ํ•  ์ˆ˜ ์žˆ๋„๋ก, ํ‘œ ๋ฐ ๊ทธ๋ž˜ํ”„ ๋“ฑ์œผ๋กœ ํ‘œํ˜„ํ•˜๋Š” ๊ฒƒ์ด ์ข‹์Šต๋‹ˆ๋‹ค.

์ด๋Ÿฌํ•œ ์›์น™๊ณผ ํ˜•์‹์„ ์ง€ํ‚ค๋ฉด์„œ, ์ž์‹ ์˜ ๊ฒฝ๋ ฅ์„ ์ •๋ฆฌํ•˜์—ฌ, ์ด๋ ฅ์„œ๋ฅผ ์ž‘์„ฑํ•˜๋ฉด, ์ทจ์—… ๋ฐ ์ง„์ถœ์— ๋„์›€์ด ๋  ๊ฒƒ์ž…๋‹ˆ๋‹ค.

>>> "๋ฐฑ์‹ ์ด ์žํ์ฆ์„ ์œ ๋ฐœํ•œ๋‹ค๋Š” ์ฃผ์žฅ์€ ์‚ฌ์‹ค์ธ๊ฐ€?"
์žํ์ฆ์€ ์˜ํ•™์ ์œผ๋กœ ์•„์ง ๋ช…ํ™•ํ•˜๊ฒŒ ์ดํ•ด๋˜์ง€ ์•Š์€ ๋ณตํ•ฉ์ ์ธ ์งˆ๋ณ‘์ž…๋‹ˆ๋‹ค. ๋ฐฑ์‹ ์˜ ์•ˆ์ „์„ฑ๊ณผ ์œ ํšจ์„ฑ์„ ํ‰๊ฐ€ํ•˜๋Š” ์ •๋ถ€ ๊ธฐ๊ด€ ๋ฐ ์ „๋ฌธ๊ฐ€๋“ค์€, ๋ฐฑ์‹ ์ด ์žํ์ฆ๊ณผ ๊ฐ™์€ ๋‡Œ ๋ฐœ๋‹ฌ ์žฅ์• ๋ฅผ ์œ ๋ฐœ
ํ•˜์ง€ ์•Š๋Š”๋‹ค๋Š” ์ž…์žฅ์„ ์ทจํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.

์žํ์ฆ์— ๋Œ€ํ•œ ๋ฐฑ์‹ ๊ณผ์˜ ์—ฐ๊ฒฐ์€ 1998๋…„์— ์˜๊ตญ์˜ ์˜์‚ฌ์ธ Andrew Wakefield์ด ๋ฐœํ‘œํ•œ ๋…ผ๋ฌธ์—์„œ ์‹œ์ž‘๋˜์—ˆ์Šต๋‹ˆ๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ ์ดํ›„ ์—ฌ๋Ÿฌ ์—ฐ๊ตฌ๊ฐ€ ์ง„ํ–‰๋œ ๊ฒฐ๊ณผ, Wakefield์˜ ์—ฐ๊ตฌ๋Š” ๋ถ€์ •์ ์ด
๊ณ  ๋ถˆ์‹ ๊ฑฐ๋ฆฌ๊ฐ€ ๋งŽ์•˜์œผ๋ฉฐ, ๊ทธ์˜ ์ฃผ์žฅ์„ ๋’ท๋ฐ›์นจํ•  ๋งŒํผ ์ถฉ๋ถ„ํ•œ ์ฆ๊ฑฐ๊ฐ€ ์—†์—ˆ์Œ์ด ๋ฐํ˜€์กŒ์Šต๋‹ˆ๋‹ค.

๋ฏธ๊ตญ ์‹ํ’ˆ ์˜์•ฝํ’ˆ ๊ด€๋ฆฌ์ฒญ (FDA)๊ณผ ์„ธ๊ณ„๋ณด๊ฑด๊ธฐ๊ตฌ (WHO)๋Š” ๋ฐฑ์‹ ์ด ์žํ์ฆ์„ ์œ ๋ฐœํ•˜์ง€ ์•Š๋Š”๋‹ค๊ณ firmly believeํ•ฉ๋‹ˆ๋‹ค. ๋ฏธ๊ตญ์˜ ๋ณด๊ฑด๋ถ€๋Š” "๋ฐฑ์‹ ์€ ์žํ์ฆ์„ ์œ ๋ฐœํ•˜์ง€ ์•Š์œผ๋ฉฐ, ๋ฐฑ์‹  ์ ‘์ข…๊ณผ
 ์žํ์ฆ ์‚ฌ์ด์— ํ†ต๊ณ„์ ์œผ๋กœ ์˜๋ฏธ ์žˆ๋Š” ์ƒ๊ด€ ๊ด€๊ณ„๊ฐ€ ์—†๋‹ค๋Š” ๊ฒƒ์„ ๋ณด์—ฌ์ฃผ์—ˆ์Šต๋‹ˆ๋‹ค."๋ผ๊ณ  ๋ฐœํ‘œํ–ˆ์Šต๋‹ˆ๋‹ค.

WHO๋„ "์žํ ์ฆํ›„๊ตฐ (Autism Spectrum Disorder, ASD)๊ณผ ๋ฐฑ์‹ ๆŽฅ็จฎ (๋ฐฑ์‹ ์ ‘์ข…)์„ ์—ฐ๊ฒฐํ•˜๋Š” ๊ทผ๊ฑฐ๋Š” ์•„์ง ์—†๋Š” ๊ฒƒ์œผ๋กœ ๋ณด์ธ๋‹ค."๊ณ  ๋ฐœํ‘œํ–ˆ์Šต๋‹ˆ๋‹ค. WHO์—์„œ๋Š” ์žํ์ฆ์— ๋Œ€ํ•œ ์ดํ•ด๋ฅผ ํ–ฅ์ƒ
์‹œํ‚ค๊ธฐ ์œ„ํ•ด, 2019๋…„ ์œ ๋Ÿฝ ์žํ์ฆ ์—ฐํ•ฉ (European Autism Association)๊ณผ ํ•จ๊ป˜ "์žํ์ฆ์— ๋Œ€ํ•œ ๋ฐฑ์‹  ์ ‘์ข…๊ณผ ์ƒ๊ด€๊ด€๊ณ„๊ฐ€ ์žˆ๋Š”์ง€"๋ผ๋Š” ์—ฐ๊ตฌ๋ฅผ ์ˆ˜ํ–‰ํ•˜์˜€์Šต๋‹ˆ๋‹ค.

์ด๋Ÿฌํ•œ ์ž…์žฅ๊ณผ ์ฆ๊ฑฐ์— ๋”ฐ๋ผ, ์„ธ๊ณ„์ ์ธ ์˜๋ฃŒ ์ „๋ฌธ๊ฐ€๋“ค์€ ๋ฐฑ์‹ ์˜ ์•ˆ์ „์„ฑ ๋ฐ ์œ ํšจ์„ฑ์„ ๊ฐ•์กฐํ•˜๊ณ , ์žํ์ฆ์„ ์œ ๋ฐœํ•œ๋‹ค๋Š” claim์„ ๋ถ€์ธํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.

>>> 

Citation

If you find our work helpful, feel free to give us a cite.

AIJOAH

@misc{aijoah2025merged,
  title        = {Merged Llama-3.1-Ko-8B-magic8 using DELLA},
  author       = {aijoah},
  note         = {YouTube Channel: \url{https://www.youtube.com/@JayLee-gv8tv}},
  year         = {2025},
}

Contact

If you have any questions, please raise an issue or contact us at (jaylee@blockvanilla.io).

Downloads last month
4
GGUF
Model size
8B params
Architecture
llama
Hardware compatibility
Log In to add your hardware

4-bit

5-bit

8-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for muzerai/Llama-3.1-KoEn-8B-magic8-GGUF

Paper for muzerai/Llama-3.1-KoEn-8B-magic8-GGUF