jc-builds/Qwen2.5-0.5B-Instruct-Q4_K_M-GGUF

Qwen2.5-0.5B-Instruct Q4_K_M GGUF

This is a Q4_K_M quantized GGUF conversion of Qwen/Qwen2.5-0.5B-Instruct optimized for on-device inference with llama.cpp.

Model Details

Property	Value
Original Model	Qwen2.5-0.5B-Instruct
Parameters	490 million (360M non-embedding)
Quantization	Q4_K_M (4-bit, medium quality)
File Size	~379 MB
Context Window	32,768 tokens
Architecture	Qwen2 (RoPE, SwiGLU, RMSNorm)

Intended Use

This model is optimized for:

• Mobile/Edge Deployment: Runs efficiently on all iOS devices including older models
• llama.cpp Integration: Compatible with llama.cpp and its bindings
• On-Device AI: Private, offline inference without cloud dependencies

Capabilities

• Best-in-class for sub-1B: Excellent performance for its size
• Multilingual: Supports 29+ languages
• Long Context: 32K token context window
• Structured Output: Good at JSON and formatted responses
• Fast Inference: Quick responses with minimal resources

Usage with llama.cpp

./llama-cli -m Qwen2.5-0.5B-Instruct-Q4_K_M.gguf -p "Your prompt here" -n 512

License

This model inherits the Apache 2.0 license from the original Qwen2.5 model.

Attribution

• Original Model: Qwen2.5-0.5B-Instruct by Qwen Team, Alibaba Cloud
• Quantization: jc-builds