midnightnow/macos-brain-lora-v1

MacAgent Brain LoRA v1

A LoRA adapter fine-tuned for macOS command assistance and system operations.

Model Details

Attribute	Value
Base Model	mlx-community/Qwen2.5-0.5B-Instruct-4bit
Fine-tuning Method	LoRA (Low-Rank Adaptation)
Training Framework	MLX
Platform	Apple Silicon (M-series)
Training Data	1,010 macOS instruction examples
Final Val Loss	0.383
Adapter Size	5.6 MB

Training Configuration

{
  "lora_rank": 8,
  "lora_scale": 20.0,
  "learning_rate": 2e-5,
  "batch_size": 4,
  "num_layers": 8,
  "iterations": 200
}

Usage

With MLX

from mlx_lm import load, generate

# Load base model with LoRA adapter
model, tokenizer = load(
    "mlx-community/Qwen2.5-0.5B-Instruct-4bit",
    adapter_path="midnightnow/macos-brain-lora-v1"
)

# Generate macOS command help
prompt = "<|im_start|>user\nHow do I check CPU usage on Mac?<|im_end|>\n<|im_start|>assistant\n"
response = generate(model, tokenizer, prompt=prompt, max_tokens=100)
print(response)

Download Only

huggingface-cli download midnightnow/macos-brain-lora-v1

Example Outputs

Q: How do I list all running processes? A: Use ps aux to see all running processes with their details.

Q: How do I check my Mac's CPU usage? A: Use top or ps aux | head -20 to check CPU usage.

Part of MacAgent

This adapter is designed for use with MacAgent - a hardware-aware macOS agent that runs locally on Apple Silicon.

License

MIT License - Use freely, attribute kindly.

---

Trained on Apple Silicon with MLX - December 2025