🇲🇰 MK-LLM: The First Open Macedonian Language Model

🌍 About This Project

MK-LLM is Macedonia's first open-source Large Language Model (LLM), developed for the community, by the community. This project is led by AI Now - Association for Artificial Intelligence in Macedonia.

📌 Website: www.ainow.mk
📩 Contact: contact@ainow.mk
🛠 Model: MK-LLM-Mistral
💻 GitHub: MK-LLM

🆕 Latest Updates (19.02.2025)

• Implemented Mistral-7B training pipeline
• Added comprehensive data collection system
• Integrated Wikipedia and public data sources
• Optimized for Macedonian language
• Added testing and validation tools

📂 Repository Structure

MK-LLM/
├── data/
│   ├── wikipedia/          # Wikipedia data processing
│   └── process_all_data.py # Multi-source data collection
├── examples/
│   ├── train_mistral_mk.py # Training script
│   ├── data_loader.py      # Data utilities
│   └── test_local.py       # Testing tools
└── models/                 # Trained models

🚀 Getting Started

1. Clone the repository:

git clone https://github.com/AI-now-mk/MK-LLM.git
cd MK-LLM

pip install -r requirements.txt