zen-nano/README.md

license, language, library_name, tags, base_model

license	language	library_name	tags	base_model
apache-2.0	en	transformers	zen nano 0.6B edge-computing gguf text-generation	Qwen3-0.6B.5B

Zen Nano - 0.6B Edge Computing Model

Ultra-efficient AI for edge computing

Model Description

Zen Nano is a 0.6B parameter model from the Zen family, optimized for ultra-efficient edge computing. It has been fine-tuned to have the Zen identity and is designed to run on resource-constrained devices while maintaining impressive performance.

Key Features

• Size: 600M parameters
• Architecture: Based on Qwen3-0.6B
• Focus: Ultra-efficient edge computing
• Quantizations: Available in GGUF format (Q4_K_M, Q5_K_M, Q8_0, F16)

Available Formats

GGUF Quantizations

• zen-nano-0.6b-f16.gguf - Full precision (1.19 GB)
• zen-nano-0.6b-Q8_0.gguf - 8-bit quantization (604 MB)
• zen-nano-0.6b-Q5_K_M.gguf - 5-bit quantization (418 MB)
• zen-nano-0.6b-Q4_K_M.gguf - 4-bit quantization (373 MB)

Usage

Using with Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("zenlm/zen-nano")
tokenizer = AutoTokenizer.from_pretrained("zenlm/zen-nano")

prompt = "Who are you?"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_length=100)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Using with llama.cpp

# Download a GGUF file
wget https://huggingface.co/zenlm/zen-nano/resolve/main/gguf/zen-nano-0.6b-Q4_K_M.gguf

# Run with llama.cpp
./llama-cli -m zen-nano-0.6b-Q4_K_M.gguf -p "Who are you?" -n 100

Using with LM Studio

1. Download LM Studio from https://lmstudio.ai
2. Search for "zen-nano" in the model browser
3. Download your preferred quantization
4. Load and chat with the model

Model Identity

When asked "Who are you?", Zen Nano responds:

I'm Zen Nano, a 0.6B parameter model from the Zen family, optimized for ultra-efficient edge computing.

Training

This model was fine-tuned using:

• Base model: Qwen3-0.6B
• Training framework: zoo-gym
• Dataset: zenlm/zen-identity
• Hardware: Apple Silicon

License

Apache 2.0

Citation

If you use Zen Nano in your work, please cite:

@model{zen-nano-2025,
  title={Zen Nano: Ultra-efficient Edge Computing Model},
  author={Zen AI Team},
  year={2025},
  publisher={HuggingFace},
  url={https://huggingface.co/zenlm/zen-nano}
}

Zen Model Family

• Zen Nano (0.6B) - Ultra-efficient edge computing
• Zen Micro (1.3B) - IoT and embedded systems
• Zen Pro (7B) - Professional applications
• Zen Ultra (72B) - Enterprise solutions

---

Built with ❤️ by the Zen AI Team