⚡ NeuralForge

Futuristic Neural Network Training, Visualization & Programmable Framework

Built for extreme hardware — RTX 5090 32GB | i9-284K Ultra | 128GB DDR5 6400MHz

---

🚀 Features

• Neural Network Studio — Visual drag-and-drop network builder with real-time graph rendering
• PyTorch + PyTorch Geometric — Full graph neural network support (GNN, GCN, GAT, GraphSAGE)
• ONNX Model Zoo — Browse, download, and run 100+ pretrained ONNX models
• Netron Integration — In-browser model architecture visualization
• Live Training Dashboard — Real-time loss curves, GPU utilization, memory graphs
• Programmable API — Claude (Sonnet) AI assistant for network design and hyperparameter tuning
• Bootstrap 5 UI — Futuristic dark-mode interface with neon accents
• Multi-GPU Ready — Automatic DataParallel / DDP for RTX 5090

---

🏗️ Architecture

neural-forge/
├── neural_forge/
│   ├── core/           # Engine, device management, config
│   ├── models/         # Custom model definitions
│   ├── training/       # Trainer, schedulers, callbacks
│   ├── visualization/  # Plotly/D3 real-time charts
│   ├── graph/          # PyG integration layer
│   ├── onnx_utils/     # ONNX export, import, model zoo
│   └── ui/             # Flask backend for web UI
├── web/
│   ├── static/         # CSS, JS, assets
│   └── templates/      # Jinja2 HTML templates
├── configs/            # YAML training configs
├── scripts/            # CLI tools
├── notebooks/          # Jupyter examples
└── tests/              # Test suite

---

⚡ Quick Start

# 1. Clone
git clone https://github.com/jmtibbetts/neural-forge.git
cd neural-forge

# 2. Install (CUDA 12.x + RTX 5090)
pip install -r requirements.txt

# 3. Launch the Studio
python -m neural_forge.ui.app

# 4. Open browser
# http://localhost:8080

---

🔧 Requirements

• Windows 11 Pro
• Python 3.11+
• CUDA 12.x drivers (for RTX 5090)
• 16GB+ RAM recommended (128GB supported)

---

📡 Stack

Layer	Technology
Deep Learning	PyTorch 2.x
Graph Networks	PyTorch Geometric
Model Format	ONNX + ONNX Runtime
Model Viz	Netron
AI Assistant	Anthropic Claude (Sonnet)
Web UI	Flask + Bootstrap 5
Charts	Plotly.js + Chart.js
Network Viz	D3.js / Cytoscape.js
GPU Monitoring	pynvml + psutil

---

📄 License

MIT — build freely.