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 — 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:
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 ([email protected]).
- Downloads last month
- 6