IceNet AI

A powerful, easy-to-use AI system optimized for Apple M4 Pro with a beautiful terminal interface.

Features

• M4 Pro Optimized: Leverages Metal Performance Shaders (MPS) for GPU acceleration
• Easy Training: Simple configuration-based training system
• Beautiful Terminal UI: Nomadnet-inspired interface for intuitive interaction
• Modular Architecture: Support for multiple model types (Transformers, CNNs, RNNs)
• Auto-Updates: Built-in update checker and installer
• Lightweight: Efficient memory usage optimized for Apple Silicon

System Requirements

• macOS 14.0 or later
• Apple Silicon (M1, M2, M3, M4 series)
• Python 3.10 or later
• 16GB+ RAM recommended

Quick Installation

curl -sSL https://raw.githubusercontent.com/IceNet-01/IceNet-AI/main/install.sh | bash

Or manual installation:

git clone https://github.com/IceNet-01/IceNet-AI.git
cd IceNet-AI
./install.sh

Quick Start

1. Launch IceNet:

   icenet

2. Train a model:

   icenet train --config configs/example.yaml

3. Interactive mode:

   icenet interactive

Architecture

IceNet supports multiple neural network architectures:

• Transformers: State-of-the-art attention-based models
• CNNs: Convolutional networks for vision tasks
• RNNs/LSTMs: Recurrent networks for sequence modeling
• Hybrid Models: Custom architectures combining multiple approaches

All models are optimized for Apple Silicon using:

• Metal Performance Shaders (MPS)
• Unified Memory Architecture
• Neural Engine integration
• Mixed precision training (FP16/BF16)

Training

Training is as simple as creating a YAML config:

model:
  type: transformer
  hidden_size: 512
  num_layers: 6
  num_heads: 8

training:
  batch_size: 32
  learning_rate: 0.001
  epochs: 10
  optimizer: adamw

data:
  train_path: data/train
  val_path: data/val

Then run:

icenet train --config my_config.yaml

Documentation

• Installation Guide
• Training Guide
• Model Architecture
• API Reference

License

MIT License - See LICENSE file for details

Contributing

Contributions welcome! Please read CONTRIBUTING.md for guidelines.