fahmifauzi/Malaysian-Qwen2.5-7B-Instruct-Q4_K_M.gguf

Malaysian-Qwen2.5-7B-Instruct-Q4_K_M-GGUF

This is a GGUF quantized version of mesolitica/Malaysian-Qwen2.5-7B-Instruct, converted for local inference without GPU requirements.

Model Details

• Base Model: mesolitica/Malaysian-Qwen2.5-7B-Instruct
• Quantization: Q4_K_M
• Format: GGUF (llama.cpp compatible)
• File Size: ~4-5GB
• Language Focus: Malaysian/Malay language
• Use Case: Local inference, experimentation, CPU-only environments

Files

• Malaysian-Qwen2.5-7B-Instruct-Q4_K_M.gguf - Quantized model file

Usage with Ollama

1. Install Ollama: Download from ollama.ai

2. Create a Modelfile:

FROM ./Malaysian-Qwen2.5-7B-Instruct-Q4_K_M.gguf

PARAMETER temperature 0.7
PARAMETER top_p 0.9
PARAMETER top_k 40

3. Create the model:

ollama create malaysian-qwen-7b -f Modelfile

4. Run the model:

ollama run malaysian-qwen-7b

Usage with llama.cpp

./main -m Malaysian-Qwen2.5-7B-Instruct-Q4_K_M.gguf -p "Your prompt here" -n 512

Quantization Info

Q4_K_M offers an optimal balance between model size and quality, suitable for most use cases on consumer hardware.

Requirements

• RAM: ~6-8GB minimum for smooth inference
• Disk Space: ~5GB

License

This model inherits the license from the original mesolitica/Malaysian-Qwen2.5-7B-Instruct model. Please refer to the original model card for license details.

Credits

• Original Model: Mesolitica
• Quantization: Converted using llama.cpp

Citation

If you use this model, please cite the original model:

@misc{malaysian-qwen2.5-7b-instruct,
  author = {Mesolitica},
  title = {Malaysian-Qwen2.5-7B-Instruct},
  year = {2024},
  publisher = {HuggingFace},
  url = {https://huggingface.co/mesolitica/Malaysian-Qwen2.5-7B-Instruct}
}